diff options
Diffstat (limited to 'intl/icu/source/test/testdata/translit_rules.txt')
| -rw-r--r-- | intl/icu/source/test/testdata/translit_rules.txt | 148 |
1 files changed, 148 insertions, 0 deletions
diff --git a/intl/icu/source/test/testdata/translit_rules.txt b/intl/icu/source/test/testdata/translit_rules.txt new file mode 100644 index 0000000000..11d12bb56b --- /dev/null +++ b/intl/icu/source/test/testdata/translit_rules.txt @@ -0,0 +1,148 @@ +# Copyright (C) 2016 and later: Unicode, Inc. and others. +# License & terms of use: http://www.unicode.org/copyright.html +#-------------------------------------------------------------------- +# Copyright (c) 1999-2004, International Business Machines +# Corporation and others. All Rights Reserved. +#-------------------------------------------------------------------- + +# InterIndic-Bengali +#:: NFD (NFC) ; +\uE001>\u0981; # SIGN CANDRABINDU +\uE002>\u0982; # SIGN ANUSVARA +\uE003>\u0983; # SIGN VISARGA +\uE005>\u0985; # LETTER A +\uE006>\u0986; # LETTER AA +\uE007>\u0987; # LETTER I +\uE008>\u0988; # LETTER II +\uE009>\u0989; # LETTER U +\uE00A>\u098A; # LETTER UU +\uE00B>\u098B; # LETTER VOCALIC R +\uE00C>\u098C; # LETTER VOCALIC L +\uE00D>\u098F; # FALLBACK +\uE00E>\u098F; # FALLBACK +\uE00F>\u098F; # LETTER E +\uE010>\u0990; # LETTER AI +\uE011>\u0993; # FALLBACK +\uE012>\u0993; # FALLBACK +\uE013>\u0993; # LETTER O +\uE014>\u0994; # LETTER AU +\uE015>\u0995; # LETTER KA +\uE016>\u0996; # LETTER KHA +\uE017>\u0997; # LETTER GA +\uE018>\u0998; # LETTER GHA +\uE019>\u0999; # LETTER NGA +\uE01A>\u099A; # LETTER CA +\uE01B>\u099B; # LETTER CHA +\uE01C>\u099C; # LETTER JA +\uE01D>\u099D; # LETTER JHA +\uE01E>\u099E; # LETTER NYA +\uE01F>\u099F; # LETTER TTA +\uE020>\u09A0; # LETTER TTHA +\uE021>\u09A1; # LETTER DDA +\uE022>\u09A2; # LETTER DDHA +\uE023>\u09A3; # LETTER NNA +\uE024>\u09A4; # LETTER TA +\uE025>\u09A5; # LETTER THA +\uE026>\u09A6; # LETTER DA +\uE027>\u09A7; # LETTER DHA +\uE028>\u09A8; # LETTER NA +\uE029>\u09A8\u09BC; # REMAP (indicExceptions.txt): \u09a9>\u09a8 = LETTER NNNA>LETTER NA +\uE02A>\u09AA; # LETTER PA +\uE02B>\u09AB; # LETTER PHA +\uE02C>\u09AC; # LETTER BA +\uE02D>\u09AD; # LETTER BHA +\uE02E>\u09AE; # LETTER MA +\uE02F>\u09AF; # LETTER YA +\uE030>\u09B0; # LETTER RA +\uE031>\u09B0\u09BC; # FALLBACK to RA +\uE032>\u09B2; # LETTER LA +\uE033>\u09B2; # REMAP (indicExceptions.txt): \u09b3>\u09b2 = LETTER LLA>LETTER LA +\uE034>\u09B2; # REMAP (indicExceptions.txt): \u09b4>\u09b2 = LETTER LLLA>LETTER LA +\uE035>\u09AC; # REMAP (indicExceptions.txt): \u09b5>\u09ac = LETTER VA>LETTER BA +\uE036>\u09B6; # LETTER SHA +\uE037>\u09B7; # LETTER SSA +\uE038>\u09B8; # LETTER SA +\uE039>\u09B9; # LETTER HA +\uE03C>\u09BC; # SIGN NUKTA +\uE03D>; # FALLBACK BLOW AWAY AVAGRAHA +\uE03E>\u09BE; # VOWEL SIGN AA +\uE03F>\u09BF; # VOWEL SIGN I +\uE040>\u09C0; # VOWEL SIGN II +\uE041>\u09C1; # VOWEL SIGN U +\uE042>\u09C2; # VOWEL SIGN UU +\uE043>\u09C3; # VOWEL SIGN VOCALIC R +\uE044>\u09C4; # VOWEL SIGN VOCALIC RR +\uE045>\u09C7; # REMAP (indicExceptions.txt): \u09c5>\u09c7 = VOWEL SIGN CANDRA E>VOWEL SIGN E +\uE046>\u09C7; # FALLBACK +\uE047>\u09C7; # VOWEL SIGN E +\uE048>\u09C8; # VOWEL SIGN AI +\uE049>\u09C7\u09BE; # REMAP (indicExceptions.txt): \u09c9>\u09cb = VOWEL SIGN CANDRA O>VOWEL SIGN O +\uE04A>\u09C7\u09BE; # FALLBACK +\uE04B>\u09C7\u09BE; # VOWEL SIGN O +\uE04C>\u09C7\u09D7; # VOWEL SIGN AU +\uE04D>\u09CD; # SIGN VIRAMA +\uE050>\u0993\u0982; # InterIndic-Bengali: OM +\ue051>; +\ue052>; +\ue053>; +\ue054>; +\uE055>; # LENGTH MARK +\uE056>\u09C8; # REMAP (indicExceptions.txt): \u09d6>\u09c8 = AI LENGTH MARK>VOWEL SIGN AI +\uE057>\u09D7; # AU LENGTH MARK +\uE058>\u0995\u09BC; # FALLBACK +\uE059>\u0996\u09BC; # REMAP (indicExceptions.txt): \u09d9>\u0996 = LETTER KHHA>LETTER KHA +\uE05A>\u0997\u09BC; # REMAP (indicExceptions.txt): \u09da>\u0997 = LETTER GHHA>LETTER GA +\uE05B>\u099C\u09BC; # REMAP (indicExceptions.txt): \u09db>\u099c = LETTER ZA>LETTER JA +\uE05C>\u09A1\u09BC; # FALLBACK +\uE05D>\u09A2\u09BC; # LETTER RHA +\uE05E>\u09AB\u09BC; # REMAP (indicExceptions.txt): \u09de>\u09ab = LETTER FA>LETTER PHA +\uE05F>\u09AF\u09BC; # LETTER YYA +\uE060>\u09E0; # LETTER VOCALIC RR +\uE061>\u09E1; # LETTER VOCALIC LL +\uE062>\u09E2; # VOWEL SIGN VOCALIC L +\uE063>\u09E3; # VOWEL SIGN VOCALIC LL +\uE064>\u0964; # DANDA +\uE065>\u0965; # DOUBLE DANDA +\uE066>\u09E6; # DIGIT ZERO +\uE067>\u09E7; # DIGIT ONE +\uE068>\u09E8; # DIGIT TWO +\uE069>\u09E9; # DIGIT THREE +\uE06A>\u09EA; # DIGIT FOUR +\uE06B>\u09EB; # DIGIT FIVE +\uE06C>\u09EC; # DIGIT SIX +\uE06D>\u09ED; # DIGIT SEVEN +\uE06E>\u09EE; # DIGIT EIGHT +\uE06F>\u09EF; # DIGIT NINE +\ue070>; # ABBREVIATION SIGN +\ue071>\u09F0; # LETTER RA WITH MIDDLE DIAGONAL +\ue072>\u09F1; # LETTER RA WITH LOWER DIAGONAL +\ue073>\u09F2; # RUPEE MARK +\ue074>\u09F3; # RUPEE SIGN +\ue075>\u09F4; # CURRENCY NUMERATOR ONE +\ue076>\u09F5; # CURRENCY NUMERATOR TWO +\ue077>\u09F6; # CURRENCY NUMERATOR THREE +\ue078>\u09F7; # CURRENCY NUMERATOR FOUR +\ue079>\u09F8; # CURRENCY NUMERATOR ONE LESS THAN THE DENOMINATOR +\ue07A>\u09F9; # CURRENCY DENOMINATOR SIXTEEN +\ue07B>\u09FA; # ISSHAR +\uE07C>; # TIPPI +\uE07D>; # ADDAK +\uE07E>; # IRI +\uE07F>; # URA +\uE080>; # EK ONKAR + +0 > \u09E6; # FALLBACK FOR TAMIL +1 > \u09E7; + + + + + + + + + + + +# :: NFC (NFD) ; +# eof |