summaryrefslogtreecommitdiffstats
path: root/man/utf8-to-roff
diff options
context:
space:
mode:
Diffstat (limited to 'man/utf8-to-roff')
-rw-r--r--man/utf8-to-roff319
1 files changed, 319 insertions, 0 deletions
diff --git a/man/utf8-to-roff b/man/utf8-to-roff
new file mode 100644
index 0000000..9bb178f
--- /dev/null
+++ b/man/utf8-to-roff
@@ -0,0 +1,319 @@
+if (true || '\'); then exec perl -CI "$0" "$@"; fi # ') {}
+# The above uses the fact that backslash isn't special in single quotes in
+# shell script, whereas in perl it escapes the following single quote.
+#
+# The problem it tries to solve is that we want perl to be run with -CI flag
+# (to have stdin interpreted as utf-8), so we would use `#! /usr/bin/perl -CI',
+# except that if we do that then perl 5.10 complains about it being too late
+# to apply -CI if the script is run with `perl -CI ./utf8-to-roff', as we want
+# to do from the Makefile. The reason we don't do `./utf8-to-roff' from the makefile
+# is that then we require the #! line to have the right location of perl instead of
+# just consulting the path. (Similarly, we could use `#! /usr/bin/env perl -CI',
+# though that still requires that /usr/bin/env exist.) The reason we don't just
+# remove the `-CI' from the #! line is that then the script couldn't be run correctly
+# with ./utf8-to-roff.
+
+
+# Converts a man page from utf8 (not understood by groff 1.18) to groff escapes.
+# I couldn't find any existing tool to convert from utf8 to groff, though I
+# seem to recall seeing some code to do so somewhere.
+#
+# Hereby released into public domain by Peter Moulder.
+use warnings;
+use strict;
+
+# Table generated automatically using:
+# zcat /usr/share/man/man7/groff_char.7.gz |groff -man -Tutf8| col -pb | grep '\\\['|
+# perl -CI -nae 'my ($ch, $seq) = @F; if (ord($ch) >= 128) { printf(" 0x\%x, q{\%s},\n", ord($ch), $seq); }'
+# with č (0x10d) manually translated as cˇ (c\[ah]). (Anyone have a better translation, e.g. using
+# overprint? \[vc] doesn't work, btw.)
+# Similarly, ć (0x107) has been manually translated as c´ (c\[aa]), and ń (0x144) as n´ (n\[aa]).
+my %map = (
+ 0xd0, q{\[-D]},
+ 0xf0, q{\[Sd]},
+ 0xde, q{\[TP]},
+ 0xfe, q{\[Tp]},
+ 0xdf, q{\[ss]},
+ 0xfb00, q{\[ff]},
+ 0xfb01, q{\[fi]},
+ 0xfb02, q{\[fl]},
+ 0xfb03, q{\[Fi]},
+ 0xfb04, q{\[Fl]},
+ 0xc6, q{\[AE]},
+ 0xe6, q{\[ae]},
+ 0x152, q{\[OE]},
+ 0x153, q{\[oe]},
+ 0x131, q{\[.i]},
+ 0xc1, q{\['A]},
+ 0xc9, q{\['E]},
+ 0xcd, q{\['I]},
+ 0xd3, q{\['O]},
+ 0xda, q{\['U]},
+ 0xdd, q{\['Y]},
+ 0xe1, q{\['a]},
+ 0xe9, q{\['e]},
+ 0xed, q{\['i]},
+ 0xf3, q{\['o]},
+ 0xfa, q{\['u]},
+ 0xfd, q{\['y]},
+ 0xc4, q{\[:A]},
+ 0xcb, q{\[:E]},
+ 0xcf, q{\[:I]},
+ 0xd6, q{\[:O]},
+ 0xdc, q{\[:U]},
+ 0x178, q{\[:Y]},
+ 0xe4, q{\[:a]},
+ 0xeb, q{\[:e]},
+ 0xef, q{\[:i]},
+ 0xf6, q{\[:o]},
+ 0xfc, q{\[:u]},
+ 0xff, q{\[:y]},
+ 0xc2, q{\[^A]},
+ 0xca, q{\[^E]},
+ 0xce, q{\[^I]},
+ 0xd4, q{\[^O]},
+ 0xdb, q{\[^U]},
+ 0xe2, q{\[^a]},
+ 0xea, q{\[^e]},
+ 0xee, q{\[^i]},
+ 0xf4, q{\[^o]},
+ 0xfb, q{\[^u]},
+ 0xc0, q{\[`A]},
+ 0xc8, q{\[`E]},
+ 0xcc, q{\[`I]},
+ 0xd2, q{\[`O]},
+ 0xd9, q{\[`U]},
+ 0xe0, q{\[`a]},
+ 0xe8, q{\[`e]},
+ 0xec, q{\[`i]},
+ 0xf2, q{\[`o]},
+ 0xf9, q{\[`u]},
+ 0xc3, q{\[~A]},
+ 0xd1, q{\[~N]},
+ 0xd5, q{\[~O]},
+ 0xe3, q{\[~a]},
+ 0xf1, q{\[~n]},
+ 0xf5, q{\[~o]},
+ 0x107, q{c\[aa]}, # Added manually; see above.
+ 0x10d, q{c\[ah]}, # Added manually; see above.
+ 0x160, q{\[vS]},
+ 0x161, q{\[vs]},
+ 0x17d, q{\[vZ]},
+ 0x17e, q{\[vz]},
+ 0xc7, q{\[,C]},
+ 0xe7, q{\[,c]},
+ 0x141, q{\[/L]},
+ 0x142, q{\[/l]},
+ 0x144, q{n\[aa]}, # Added manually; see above.
+ 0xd8, q{\[/O]},
+ 0xf8, q{\[/o]},
+ 0xc5, q{\[oA]},
+ 0xe5, q{\[oa]},
+ 0x2dd, q{\[a"]},
+ 0xaf, q{\[a-]},
+ 0x2d9, q{\[a.]},
+ 0xb4, q{\[aa]},
+ 0x2d8, q{\[ab]},
+ 0xb8, q{\[ac]},
+ 0xa8, q{\[ad]},
+ 0x2c7, q{\[ah]},
+ 0x2da, q{\[ao]},
+ 0x2db, q{\[ho]},
+ 0x223c, q{\[ti]},
+ 0x201e, q{\[Bq]},
+ 0x201a, q{\[bq]},
+ 0x201c, q{\[lq]},
+ 0x201d, q{\[rq]},
+ 0x2018, q{\[oq]},
+ 0x2019, q{\[cq]},
+ 0xab, q{\[Fo]},
+ 0xbb, q{\[Fc]},
+ 0x2039, q{\[fo]},
+ 0x203a, q{\[fc]},
+ 0xa1, q{\[r!]},
+ 0xbf, q{\[r?]},
+ 0x2014, q{\[em]},
+ 0x2013, q{\[en]},
+ 0x2010, q{\[hy]},
+ 0x2329, q{\[la]},
+ 0x232a, q{\[ra]},
+ 0x2190, q{\[<-]},
+ 0x2192, q{\[->]},
+ 0x2194, q{\[<>]},
+ 0x2193, q{\[da]},
+ 0x21d1, q{\[ua]},
+ 0x21d0, q{\[lA]},
+ 0x21d2, q{\[rA]},
+ 0x21d4, q{\[hA]},
+ 0x21d3, q{\[dA]},
+ 0x21d1, q{\[uA]},
+ 0x2500, q{\[an]},
+ 0x2502, q{\[br]},
+ 0x2502, q{\[bv]},
+ 0xa6, q{\[bb]},
+ 0x25ef, q{\[ci]},
+ 0xb7, q{\[bu]},
+ 0x2021, q{\[dd]},
+ 0x2020, q{\[dg]},
+ 0x25ca, q{\[lz]},
+ 0x25a1, q{\[sq]},
+ 0xb6, q{\[ps]},
+ 0xa7, q{\[sc]},
+ 0x261c, q{\[lh]},
+ 0x261e, q{\[rh]},
+ 0x240d, q{\[CR]},
+ 0xa9, q{\[co]},
+ 0xae, q{\[rg]},
+ 0x2122, q{\[tm]},
+ 0x21d1, q{\[Do]},
+ 0xa2, q{\[ct]},
+ 0x20ac, q{\[eu]},
+ 0x20ac, q{\[Eu]},
+ 0xa5, q{\[Ye]},
+ 0xa3, q{\[Po]},
+ 0xa4, q{\[Cs]},
+ 0x192, q{\[Fn]},
+ 0xb0, q{\[de]},
+ 0x2030, q{\[%0]},
+ 0x2032, q{\[fm]},
+ 0x2033, q{\[sd]},
+ 0xb5, q{\[mc]},
+ 0xaa, q{\[Of]},
+ 0xba, q{\[Om]},
+ 0x2227, q{\[AN]},
+ 0x2228, q{\[OR]},
+ 0xac, q{\[no]},
+ 0x2203, q{\[te]},
+ 0x2200, q{\[fa]},
+ 0x220b, q{\[st]},
+ 0x2234, q{\[3d]},
+ 0x2234, q{\[tf]},
+ 0xbd, q{\[12]},
+ 0xbc, q{\[14]},
+ 0xbe, q{\[34]},
+ 0xb9, q{\[S1]},
+ 0xb2, q{\[S2]},
+ 0xb3, q{\[S3]},
+ 0xb1, q{\[+-]},
+ 0xb1, q{\[t+-]},
+ 0xb7, q{\[pc]},
+ 0x22c5, q{\[md]},
+ 0xd7, q{\[mu]},
+ 0xd7, q{\[tmu]},
+ 0x2297, q{\[c*]},
+ 0x2295, q{\[c+]},
+ 0xf7, q{\[di]},
+ 0xf7, q{\[tdi]},
+ 0x2044, q{\[f/]},
+ 0x2217, q{\[**]},
+ 0x2264, q{\[<=]},
+ 0x2265, q{\[>=]},
+ 0x2260, q{\[!=]},
+ 0x2261, q{\[==]},
+ 0x2245, q{\[=~]},
+ 0x223c, q{\[ap]},
+ 0x2248, q{\[~~]},
+ 0x2248, q{\[~=]},
+ 0x221d, q{\[pt]},
+ 0x2205, q{\[es]},
+ 0x2208, q{\[mo]},
+ 0x2209, q{\[nm]},
+ 0x2284, q{\[nb]},
+ 0x2282, q{\[sb]},
+ 0x2283, q{\[sp]},
+ 0x2286, q{\[ib]},
+ 0x2287, q{\[ip]},
+ 0x2229, q{\[ca]},
+ 0x222a, q{\[cu]},
+ 0x2220, q{\[/_]},
+ 0x22a5, q{\[pp]},
+ 0x222b, q{\[is]},
+ 0x2211, q{\[sum]},
+ 0x220f, q{\[product]},
+ 0x2207, q{\[gr]},
+ 0x221a, q{\[sr]},
+ 0x203e, q{\[rn]},
+ 0x221e, q{\[if]},
+ 0x2135, q{\[Ah]},
+ 0x2111, q{\[Im]},
+ 0x211c, q{\[Re]},
+ 0x2118, q{\[wp]},
+ 0x2202, q{\[pd]},
+ 0x391, q{\[*A]},
+ 0x392, q{\[*B]},
+ 0x39e, q{\[*C]},
+ 0x394, q{\[*D]},
+ 0x395, q{\[*E]},
+ 0x3a6, q{\[*F]},
+ 0x393, q{\[*G]},
+ 0x398, q{\[*H]},
+ 0x399, q{\[*I]},
+ 0x39a, q{\[*K]},
+ 0x39b, q{\[*L]},
+ 0x39c, q{\[*M]},
+ 0x39d, q{\[*N]},
+ 0x39f, q{\[*O]},
+ 0x3a0, q{\[*P]},
+ 0x3a8, q{\[*Q]},
+ 0x3a1, q{\[*R]},
+ 0x3a3, q{\[*S]},
+ 0x3a4, q{\[*T]},
+ 0x3a5, q{\[*U]},
+ 0x3a9, q{\[*W]},
+ 0x3a7, q{\[*X]},
+ 0x397, q{\[*Y]},
+ 0x396, q{\[*Z]},
+ 0x3b1, q{\[*a]},
+ 0x3b2, q{\[*b]},
+ 0x3be, q{\[*c]},
+ 0x3b4, q{\[*d]},
+ 0x3b5, q{\[*e]},
+ 0x3c6, q{\[*f]},
+ 0x3d5, q{\[+f]},
+ 0x3b3, q{\[*g]},
+ 0x3b8, q{\[*h]},
+ 0x3d1, q{\[+h]},
+ 0x3b9, q{\[*i]},
+ 0x3ba, q{\[*k]},
+ 0x3bb, q{\[*l]},
+ 0x3bc, q{\[*m]},
+ 0x3bd, q{\[*n]},
+ 0x3bf, q{\[*o]},
+ 0x3c0, q{\[*p]},
+ 0x3d6, q{\[+p]},
+ 0x3c8, q{\[*q]},
+ 0x3c1, q{\[*r]},
+ 0x3c3, q{\[*s]},
+ 0x3c4, q{\[*t]},
+ 0x3c5, q{\[*u]},
+ 0x3c9, q{\[*w]},
+ 0x3c7, q{\[*x]},
+ 0x3b7, q{\[*y]},
+ 0x3b6, q{\[*z]},
+ 0x3c2, q{\[ts]},
+ 0x2663, q{\[CL]},
+ 0x2660, q{\[SP]},
+ 0x2665, q{\[HE]},
+ 0x2666, q{\[DI]},
+);
+
+#while(<>) {
+# s/([^ -~])/(ord($1) < 128 ? $1 : defined($map{$1}) ? $map{$1} : sprintf("\\u%4x", $1))/ge;
+# print;
+#}
+#exit 0;
+
+my $ch;
+while(defined($ch = getc(STDIN))) {
+ my $ord = ord($ch);
+ if ($ord < 128) {
+ print $ch;
+ } else {
+ my $out = $map{$ord};
+ if (defined($out)) {
+ print $out;
+ } else {
+ die "Untranslatable character \\u" . sprintf("%X", ord($ch)) . " / `$ch'";
+ }
+ }
+}