From: imiyazaki Date: Sun, 2 Nov 2003 14:31:12 +0000 (+0000) Subject: not maintained. X-Git-Url: http://git.chise.org/gitweb/?a=commitdiff_plain;h=f949214226069e6b0d5c83d4bbf286948bd651e9;p=chise%2Fomega.git not maintained. --- diff --git a/chise2otf/chise2otf-5.6 b/chise2otf/chise2otf-5.6 new file mode 100755 index 0000000..f855ea8 --- /dev/null +++ b/chise2otf/chise2otf-5.6 @@ -0,0 +1,826 @@ +#!/usr/bin/perl + +use strict; +use vars qw($opt_in_cs $opt_order $opt_kage $opt_replace + $opt_help $usage + $in_cs $out_cs $i @chars + @order $order %order + @texmacro + $char $char_id $out_char + $ids $ids_argc %ids $idsdb + $idsdata_file $ids_start $font_start + $perl56 $perl58 + @CDP @HZK @GT + ); +use Getopt::Long; +use utf8; +use Fcntl ':flock'; +use Chise_utils ':all'; + +my $omegadb_path="/usr/local/lib/chise/omega"; +$omegadb_path=~s!/$!!; + +my $makefonts="$omegadb_path/makefonts.pl"; +my $exec_makefonts=0; +my $geta=pack("S",8750|0x8080); + +if($^V and $^V ge v5.8){ + $perl58=1; +}elsif($^V and $^V ge v5.6){ + $perl56=1; +}else{ + print STDERR "This versin is not supported."; +} +if($perl58){ + eval "use Encode"; + binmode(STDIN, ':encoding(utf8)'); + binmode(STDOUT, ':encoding(utf8)'); +} + +&GetOptions("in=s"=>\$opt_in_cs, + "order=s"=>\$opt_order, + "replace",\$opt_replace, + "kage",\$opt_kage, + "help",\$opt_help); + +$usage=<] [-o ] [-k] + input coding system: (default: ucs\@mcs) + ucs\@mcs, ucs\@cns, ucs\@gb, ucs\@jis, ucs\@ks + order of kanji: (default: j) + c: CNS + g: GB + j: JIS + k: KS + G: GT + m: Multi, use \\UTFM of otf.sty + You can also combine them, ex. jtcgkm + k: use Kage server. + r: replace r and l with dot below to those with circle below. +EOF + +%order=('c'=>'UniCNS', + 'g'=>'UniGB', + 'j'=>'UniJIS', + 'k'=>'UniKS', + 'G'=>'GT', + 'm'=>'Multi', + ); + +if($opt_help){ + print STDERR $usage; + exit 1; +} + +if(not defined($opt_in_cs)){ + print STDERR "Assuming input coding system is ucs\@mcs.\n"; + $in_cs='ucs@mcs'; +}else{ + if(($opt_in_cs eq 'ucs@mcs')or + ($opt_in_cs eq 'ucs@cns')or + ($opt_in_cs eq 'ucs@gb')or + ($opt_in_cs eq 'ucs@jis')or + ($opt_in_cs eq 'ucs@ks')){ + $in_cs=$opt_in_cs; + }else{ + print STDERR "Invalid arguments.\n"; + print STDERR $usage; + exit 1; + } +} + +if(not defined($opt_order)){ + print STDERR "Using only UniJIS.\n"; + $opt_order='j'; +} + +if(defined($opt_order)){ + if($opt_order=~/^[cgjkGm]+$/){ + @order=split(//,$opt_order); + @order=map {$order{$_}} @order; + }else{ + print STDERR "Invalid option!\n"; + print STDERR $usage; + exit 1; + } +} + +if($opt_kage){ + $exec_makefonts=1; +} + +&setup_texmacro(); +if($opt_replace){ + $texmacro[0x1E36]='{\ifmmode \mbox{\ucirc{L}}\else \ucirc{L}\fi}'; + $texmacro[0x1E37]='{\ifmmode \mbox{\ucirc{l}}\else \ucirc{l}\fi}'; + $texmacro[0x1E38]='{\ifmmode \ucirc{\bar{L}}\else \ucirc{\={L}}\fi}'; + $texmacro[0x1E39]='{\ifmmode \ucirc{\bar{l}}\else \ucirc{\={l}}\fi}'; + $texmacro[0x1E5A]='{\ifmmode \mbox{\ucirc{R}}\else \ucirc{R}\fi}'; + $texmacro[0x1E5B]='{\ifmmode \mbox{\ucirc{r}}\else \ucirc{r}\fi}'; + $texmacro[0x1E5C]='{\ifmmode \ucirc{\bar{R}}\else \ucirc{\={R}}\fi}'; + $texmacro[0x1E5D]='{\ifmmode \ucirc{\bar{r}}\else \ucirc{\={r}}\fi}'; +} + +$idsdata_file="$omegadb_path/idsdata.pl"; +$ids_start=0x00; +$font_start=0; + +if(-e $idsdata_file){ + open(IDSDATA,"+<$idsdata_file") or die; + flock(IDSDATA,LOCK_EX); + seek(IDSDATA,0,0); + while(){ + eval $_; + } + seek(IDSDATA,0,0); +# require $idsdata_file; +}else{ + open(IDSDATA,">$idsdata_file") or die; + flock(IDSDATA,LOCK_EX); + seek(IDSDATA,0,0); +} + +$ids_argc=0; +$ids=""; + +@GT=(#"=gt","=gt-k", + "=gt-pj-1","=gt-pj-2","=gt-pj-3","=gt-pj-4","=gt-pj-5", + "=gt-pj-6","=gt-pj-7","=gt-pj-8","=gt-pj-9","=gt-pj-10", + "=gt-pj-11" + #,"=gt-pj-k1","=gt-pj-k2" + ); +@HZK=("=hanziku-1","=hanziku-2","=hanziku-3","=hanziku-4", + "=hanziku-5","=hanziku-6","=hanziku-7","=hanziku-8", + "=hanziku-9","=hanziku-10","=hanziku-11","=hanziku-12"); +@CDP=("=big5-cdp"); + +while(<>){ + if($in_cs ne 'ucs@mcs'){ + s/(.)/pack("U",&get_char_id(unpack("U",$1),$in_cs))/ge; + } + s/((?:^|[^\\])(?:\\\\)*)(&.*?;)/$1.&de_er($2)/ge; + s/((?:^|[^\\])(?:\\\\)*)\\([$idc])/$1.'\UTFM{'.sprintf("%X",unpack("U",$2)).'}'/ge; + @chars=split(//); + for($i=0;$i<=$#chars;$i++){ + $char=$chars[$i]; + $char_id=unpack("U",$chars[$i]); + + if($char_id<=0x7f){ + # Basic Latin + print $char; + next; + }elsif(defined($texmacro[$char_id]) and $texmacro[$char_id]){ + # already defined for + # 0080..00FF; Latin-1 Supplement + # 0100..017F; Latin Extended-A + # 1E00..1EFF; Latin Extended Additional + print $texmacro[$char_id]; + next; + }elsif($char_id>=0x2ff0 and $char_id<=0x2fff){ + # Ideographic Description Characters + print &ids_parse(); + next; + }else{ + if(($out_char=&get_output_char($char_id))){ + print $out_char; + }elsif($char_id >= 0x20000 && $char_id <=0x2a6df){ + # CJK Unified Ideographs Extension B + if(not defined($ids{$char}) and $ids{$char}[1]>=0){ + $ids{$char}[0]=$font_start; + $ids{$char}[1]=$ids_start; + $ids_start++; + if($ids_start>255){ + $ids_start=0; + $font_start++; + } + } + print "{\\fontencoding{OT1}\\fontfamily{" . + sprintf("chise%03d",$ids{$char}[0]) . + "}\\selectfont\\char$ids{$char}[1]}"; + next; + }else{ + print &get_macro_for_ids(&get_ids($char)); + } + } + } +} + +print IDSDATA 'use utf8;',"\n"; +foreach $ids (keys %ids){ + print IDSDATA '$ids{\'',$ids,'\'}=' + ,'[',join ",",@{$ids{$ids}},"];\n" if($perl56); + print IDSDATA '$ids{\'',encode('utf8',$ids),'\'}=' + ,'[',join ",",@{$ids{$ids}},"];\n" if($perl58); +} +print IDSDATA '$font_start=',$font_start,";\n"; +print IDSDATA '$ids_start=',$ids_start,";\n"; +print IDSDATA "1;"; + +if($exec_makefonts){ + if(-f $makefonts){ + system("/usr/bin/env","perl",$makefonts) ==0 + or die "Could not make fonts: $?"; + }else{ + print STDERR "cannot find $makefonts\n"; + } +} + +exit 0; + +sub ids_parse{ + # arguments: none + # return: return character for output, + # TeX macro for ids, + # or GETA character if ids is invalid. + my($ids,$ids_argc)=&ids_rest("",0,$chars[$i]); + + while($ids_argc>0){ + # We are in IDS. + $i++; + if($i>$#chars){ + print STDERR "IDS parse error: $ids\n"; + return $geta; + } + + ($ids,$ids_argc)=&ids_rest($ids,$ids_argc,$chars[$i]); + if($ids_argc==0){ + if(($char_id=&get_char_id_for_ids($ids)) + and($out_char=&get_output_char($char_id))){ + return $out_char; + }else{ + return &get_macro_for_ids($ids) if($perl56); + return encode('utf8', &get_macro_for_ids($ids)) if($perl58); + } + } + } +} + +sub ids_rest{ + # arguments: , , + # return: ids and rest number of arguments for ids. + my($ids,$ids_argc,$char)=@_; + my($argc); + $argc=&ids_argc($char); + if($argc){ + $ids_argc+= $ids_argc==0 ? $argc : $argc-1; + }else{ + $ids_argc--; + } + $ids.=$char if($perl56); + $ids.=encode('utf8',$char) if($perl58); + return ($ids,$ids_argc); +} + +sub get_macro_for_ids{ + # argument: + # return: TeX macro for ids + # or GETA character if ids is invalid for KAGE. + my($ids)=@_; +# return $geta if(not $exec_makefonts); + $ids=&normalize_ids($ids,"UniJIS"); + return $geta if(($ids!~/[$idc]/) + or($ids=~/[\x{10000}-]/)); + #irregular for KAGE. + if(not defined($ids{$ids}) and $ids{$ids}[1]>=0){ + $ids{$ids}[0]=$font_start; + $ids{$ids}[1]=$ids_start; + $ids_start++; + } + if($ids_start>255){ + $ids_start=0; + $font_start++; + } + return "{\\fontencoding{T1}\\fontfamily{" + .sprintf("chisesub%03d",$ids{$ids}[0]) + ."}\\selectfont\\char$ids{$ids}[1]}"; +} + +sub normalize_ids{ + # argument: , + # return: ids or GETA character if ids is invalid for KAGE. + my($ids,$out_cs)=@_; + $ids = decode('utf8', $ids) if $perl58; + $out_cs=~s/Uni(.+)/'ucs@'.lc($1)/e; + + my $output_ids=""; + my($char,$char_id,$output_char_id); + while($ids=~m/(.)/g){ + $char=$1; + $char_id=unpack("U",$char); + if($char=~/[$idc]/){ + $output_ids.=$char; + }elsif($output_char_id=&get_char_attribute($char,"=$out_cs")){ + $output_ids.=pack("U",$output_char_id); + }elsif($output_char_id=&get_char_attribute($char,"=ucs")){ + $output_ids.=pack("U",$output_char_id); + }else{ + return $geta; + } + } + return $output_ids; +} + +sub get_output_char{ + # argument: + # return: character in EUC-JP or TeX macro for pTeX. + my($char_id)=@_; + my($char,$out_char,$out_char_id,$gt); + + $char=pack('U',$char_id); + + if($out_char=&get_char_attribute($char,'=jis-x0208')){ + return pack("S",$out_char|0x8080); + }else{ + foreach $out_cs (@order){ + if($out_cs eq 'UniJIS' + and &get_char_attribute($char,"vnd-adobe-cid-unijis-utf16-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@jis') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@jis') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return "\\UTF{".sprintf("%X",$out_char_id)."}"; + } + }elsif($out_cs eq 'UniGB' + and &get_char_attribute($char,"vnd-adobe-cid-unigb-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@gb') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@gb') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return "\\UTFC{".sprintf("%X",$out_char_id)."}"; + } + }elsif($out_cs eq 'UniCNS' + and &get_char_attribute($char,"vnd-adobe-cid-unicns-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@cns') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@cns') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return "\\UTFT{".sprintf("%X",$out_char_id)."}"; + } + }elsif($out_cs eq 'UniKS' + and &get_char_attribute($char,"vnd-adobe-cid-uniks-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@ks') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@ks') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return "\\UTFK{".sprintf("%X",$out_char_id)."}"; + } + }elsif($out_cs eq 'GT'){ + return $gt if($gt=&get_macro_for_GT($char_id)); + }elsif($out_cs eq 'Multi'){ + if($out_char_id=&get_char_attribute($char,'=ucs')){ + return "\\UTFM{".sprintf("%X",$out_char_id)."}"; + } + } + } + } + return undef; +} + +sub get_ids{ + # argument: + # return: ids + my($char)=@_; + my $ids=""; + $ids=&get_char_attribute($char,"ids-aggregated") + or &get_char_attribute($char,"ids"); + $ids=decode('utf8', $ids) if($perl58); + return $ids; +} + +sub get_char_id_for_ids{ + # argument: + # return: char-id + my($ids)=@_; + my($output_char); + $ids=decode('utf8', $ids) if($perl58); + + if(($output_char)=&get_chars_matching("ids",$ids)){ + return unpack("U",$output_char); + }else{ + return undef; + } +} + +sub get_char_id{ + # argument: , + # return: char-id. + my($char_id,$in_cs)=@_; + my($output_char); + + if(($output_char)=&get_chars_matching("=$in_cs",$char_id)){ + return unpack("U",$output_char); + }else{ + return $char_id; + } +} + +sub get_macro_for_GT{ + # argument: + # return: TeX macro for GT fonts. + my($char_id)=@_; + my($char,$gt,$GT); + $char=pack("U",$char_id); + foreach (@GT){ + if($gt=&get_char_attribute($char,$_)){ + m/gt\-pj\-(\d+)/ and $GT=$1; + last; + } + } + if($gt){ + return "{\\fontfamily{" + .sprintf("gto%02d",$GT) + ."}\\selectfont\\char".($gt|0x8080)."}"; + }else{ + return undef; + } +} + +sub setup_texmacro{ + # Most of the TeX macros below are derived from ucharacters.sty of + # jadeTeX 3.13. To use all the macros, you need unicode.sty and + # dummyels.sty of jadeTeX and extaccent.sty of CHISE. + $texmacro[0xffff]=""; + # 0080..00FF; Latin-1 Supplement + $texmacro[0xA0]='{\ensuremath{\penalty10000\ }}'; + $texmacro[0xA1]='{\textexclamdown}'; + $texmacro[0xA2]='{\ifmmode \mbox{\textcent}\else \textcent \fi}'; + $texmacro[0xA3]='{\ifmmode \pounds \else \textsterling \fi}'; + $texmacro[0xA4]='{\ifmmode \mbox{\textcurrency}\else \textcurrency \fi}'; + $texmacro[0xA5]='{\ifmmode \yen \else \textyen \fi}'; + $texmacro[0xA6]='{\textbrokenbar}'; + $texmacro[0xA7]='{\ifmmode \S \else \textsection \fi}'; + $texmacro[0xA8]='{\textasciidieresis}'; + $texmacro[0xA9]='{\ifmmode \copyright \else \textcopyright \fi}'; + $texmacro[0xAA]='{\textordfeminine}'; + $texmacro[0xAB]='{\ifmmode \mbox{\guillemotleft}\else \guillemotleft \fi}'; + $texmacro[0xAC]='{\ensuremath{\lnot }}'; + $texmacro[0xAD]='{\ensuremath{\-}}'; + $texmacro[0xAE]='{\ifmmode \circledR \else \textregistered \fi}'; + $texmacro[0xAF]='{\textasciimacron}'; + $texmacro[0xB0]='{\ifmmode \mbox{\textdegree}\else \textdegree \fi}'; + $texmacro[0xB1]='{\ensuremath{\pm }}'; + $texmacro[0xB2]='{\ensuremath{{^2}}}'; + $texmacro[0xB3]='{\ensuremath{{^3}}}'; + $texmacro[0xB4]='{\textasciiacute}'; + $texmacro[0xB5]='{\ensuremath{\mathrm{\mu}}}'; + $texmacro[0xB6]='{\ifmmode \P \else \textparagraph \fi}'; + $texmacro[0xB7]='{\ensuremath{\cdot }}'; + $texmacro[0xB8]='{\ifmmode \mbox{\c{}}\else \c{}\fi}'; + $texmacro[0xB9]='{\ensuremath{{^1}}}'; + $texmacro[0xBA]='{\textordmasculine}'; + $texmacro[0xBB]='{\ifmmode \mbox{\guillemotright}\else \guillemotright \fi}'; + $texmacro[0xBC]='{\textonequarter}'; + $texmacro[0xBD]='{\textonehalf}'; + $texmacro[0xBE]='{\textthreequarters}'; + $texmacro[0xBF]='{\textquestiondown}'; + $texmacro[0xC0]='{\ifmmode \grave{A}\else \`{A}\fi}'; + $texmacro[0xC1]='{\ifmmode \acute{A}\else \\\'{A}\fi}'; + $texmacro[0xC2]='{\ifmmode \hat{A}\else \^{A}\fi}'; + $texmacro[0xC3]='{\ifmmode \tilde{A}\else \~{A}\fi}'; + $texmacro[0xC4]='{\ifmmode \ddot{A}\else \"{A}\fi}'; + $texmacro[0xC5]='{\AA}'; + $texmacro[0xC6]='{\AE}'; + $texmacro[0xC7]='{\ifmmode \mbox{\c{C}}\else \c{C}\fi}'; + $texmacro[0xC8]='{\ifmmode \grave{E}\else \`{E}\fi}'; + $texmacro[0xC9]='{\ifmmode \acute{E}\else \\\'{E}\fi}'; + $texmacro[0xCA]='{\ifmmode \hat{E}\else \^{E}\fi}'; + $texmacro[0xCB]='{\ifmmode \ddot{E}\else \"{E}\fi}'; + $texmacro[0xCC]='{\ifmmode \grave{I}\else \`{I}\fi}'; + $texmacro[0xCD]='{\ifmmode \acute{I}\else \\\'{I}\fi}'; + $texmacro[0xCE]='{\ifmmode \hat{I}\else \^{I}\fi}'; + $texmacro[0xCF]='{\ifmmode \ddot{I}\else \"{I}\fi}'; + $texmacro[0xD0]='{\DH}'; + $texmacro[0xD1]='{\ifmmode \tilde{N}\else \~{N}\fi}'; + $texmacro[0xD2]='{\ifmmode \grave{O}\else \`{O}\fi}'; + $texmacro[0xD3]='{\ifmmode \acute{O}\else \\\'{O}\fi}'; + $texmacro[0xD4]='{\ifmmode \hat{O}\else \^{O}\fi}'; + $texmacro[0xD5]='{\ifmmode \tilde{O}\else \~{O}\fi}'; + $texmacro[0xD6]='{\ifmmode \ddot{O}\else \"{O}\fi}'; + $texmacro[0xD7]='{\ifmmode \times \else \texttimes \fi}'; + $texmacro[0xD8]='{\O}'; + $texmacro[0xD9]='{\ifmmode \grave{U}\else \`{U}\fi}'; + $texmacro[0xDA]='{\ifmmode \acute{U}\else \\\'{U}\fi}'; + $texmacro[0xDB]='{\ifmmode \hat{U}\else \^{U}\fi}'; + $texmacro[0xDC]='{\ifmmode \ddot{U}\else \"{U}\fi}'; + $texmacro[0xDD]='{\ifmmode \acute{Y}\else \\\'{Y}\fi}'; + $texmacro[0xDE]='{\TH}'; + $texmacro[0xDF]='{\ss}'; + $texmacro[0xE0]='{\ifmmode \grave{a}\else \`{a}\fi}'; + $texmacro[0xE1]='{\ifmmode \acute{a}\else \\\'{a}\fi}'; + $texmacro[0xE2]='{\ifmmode \hat{a}\else \^{a}\fi}'; + $texmacro[0xE3]='{\ifmmode \tilde{a}\else \~{a}\fi}'; + $texmacro[0xE4]='{\ifmmode \ddot{a}\else \"{a}\fi}'; + $texmacro[0xE5]='{\aa}'; + $texmacro[0xE6]='{\ae}'; + $texmacro[0xE7]='{\ifmmode \mbox{\c{c}}\else \c{c}\fi}'; + $texmacro[0xE8]='{\ifmmode \grave{e}\else \`{e}\fi}'; + $texmacro[0xE9]='{\ifmmode \acute{e}\else \\\'{e}\fi}'; + $texmacro[0xEA]='{\ifmmode \hat{e}\else \^{e}\fi}'; + $texmacro[0xEB]='{\ifmmode \ddot{e}\else \"{e}\fi}'; + $texmacro[0xEC]='{\ifmmode \grave{\imath}\else \`{\i}\fi}'; + $texmacro[0xED]='{\ifmmode \acute{\imath}\else \\\'{\i}\fi}'; + $texmacro[0xEE]='{\ifmmode \hat{\imath}\else \^{\i}\fi}'; + $texmacro[0xEF]='{\ifmmode \ddot{\imath}\else \"{\i}\fi}'; + $texmacro[0xF0]='{\dh}'; + $texmacro[0xF1]='{\ifmmode \tilde{n}\else \~{n}\fi}'; + $texmacro[0xF2]='{\ifmmode \grave{o}\else \`{o}\fi}'; + $texmacro[0xF3]='{\ifmmode \acute{o}\else \\\'{o}\fi}'; + $texmacro[0xF4]='{\ifmmode \hat{o}\else \^{o}\fi}'; + $texmacro[0xF5]='{\ifmmode \tilde{o}\else \~{o}\fi}'; + $texmacro[0xF6]='{\ifmmode \ddot{o}\else \"{o}\fi}'; + $texmacro[0xF7]='{\ensuremath{\div }}'; + $texmacro[0xF8]='{\o}'; + $texmacro[0xF9]='{\ifmmode \grave{u}\else \`{u}\fi}'; + $texmacro[0xFA]='{\ifmmode \acute{u}\else \\\'{u}\fi}'; + $texmacro[0xFB]='{\ifmmode \hat{u}\else \^{u}\fi}'; + $texmacro[0xFC]='{\ifmmode \ddot{u}\else \"{u}\fi}'; + $texmacro[0xFD]='{\ifmmode \acute{y}\else \\\'{y}\fi}'; + $texmacro[0xFE]='{\th}'; + $texmacro[0xFF]='{\ifmmode \ddot{y}\else \"{y}\fi}'; + # 0100..017F; Latin Extended-A + $texmacro[0x100]='{\ifmmode \bar{A}\else \={A}\fi}'; + $texmacro[0x101]='{\ifmmode \bar{a}\else \={a}\fi}'; + $texmacro[0x102]='{\ifmmode \breve{A}\else \u{A}\fi}'; + $texmacro[0x103]='{\ifmmode \u{a}\else \u{a}\fi}'; + $texmacro[0x104]='{\k{A}}'; + $texmacro[0x105]='{\k{a}}'; + $texmacro[0x106]='{\ifmmode \acute{C}\else \\\'{C}\fi}'; + $texmacro[0x107]='{\ifmmode \acute{c}\else \\\'{c}\fi}'; + $texmacro[0x108]='{\ifmmode \hat{C}\else \^{C}\fi}'; + $texmacro[0x109]='{\ifmmode \hat{c}\else \^{c}\fi}'; + $texmacro[0x10A]='{\ifmmode \dot{C}\else \.{C}\fi}'; + $texmacro[0x10B]='{\ifmmode \dot{c}\else \.{c}\fi}'; + $texmacro[0x10C]='{\ifmmode \check{C}\else \v{C}\fi}'; + $texmacro[0x10D]='{\ifmmode \check{c}\else \v{c}\fi}'; + $texmacro[0x10E]='{\ifmmode \check{D}\else \v{D}\fi}'; + $texmacro[0x10F]='{\ifmmode \check{d}\else \v{d}\fi}'; + $texmacro[0x110]='{\DJ}'; + $texmacro[0x111]='{\dj}'; + $texmacro[0x112]='{\ifmmode \bar{E}\else \={E}\fi}'; + $texmacro[0x113]='{\ifmmode \bar{e}\else \={e}\fi}'; + $texmacro[0x114]='{\ifmmode \breve{E}\else \u{E}\fi}'; + $texmacro[0x115]='{\ifmmode \breve{e}\else \u{e}\fi}'; + $texmacro[0x116]='{\ifmmode \dot{E}\else \.{E}\fi}'; + $texmacro[0x117]='{\ifmmode \dot{e}\else \.{e}\fi}'; + $texmacro[0x118]='{\ifmmode \k{E}\else \k{E}\fi}'; + $texmacro[0x119]='{\k{e}}'; + $texmacro[0x11A]='{\ifmmode \check{E}\else \v{E}\fi}'; + $texmacro[0x11B]='{\ifmmode \check{e}\else \v{e}\fi}'; + $texmacro[0x11C]='{\ifmmode \hat{G}\else \^{G}\fi}'; + $texmacro[0x11D]='{\ifmmode \hat{g}\else \^{g}\fi}'; + $texmacro[0x11E]='{\ifmmode \breve{G}\else \u{G}\fi}'; + $texmacro[0x11F]='{\ifmmode \breve{g}\else \u{g}\fi}'; + $texmacro[0x120]='{\ifmmode \dot{G}\else \.{G}\fi}'; + $texmacro[0x121]='{\ifmmode \dot{g}\else \.{g}\fi}'; + $texmacro[0x122]='{\ifmmode \mbox{\c{G}}\else \c{G}\fi}'; + $texmacro[0x123]='{\ifmmode \mbox{\c{g}}\else \c{g}\fi}'; + $texmacro[0x124]='{\ifmmode \hat{H}\else \^{H}\fi}'; + $texmacro[0x125]='{\ifmmode \hat{h}\else \^{h}\fi}'; + $texmacro[0x126]='{{\fontencoding{LELA}\selectfont\char40}}'; + $texmacro[0x128]='{\ifmmode \tilde{I}\else \~{I}\fi}'; + $texmacro[0x129]='{\ifmmode \tilde{\imath}\else \~{\i}\fi}'; + $texmacro[0x12A]='{\ifmmode \bar{I}\else \={I}\fi}'; + $texmacro[0x12B]='{\ifmmode \bar{\imath}\else \={\i}\fi}'; + $texmacro[0x12C]='{\ifmmode \breve{I}\else \u{I}\fi}'; + $texmacro[0x12D]='{\ifmmode \breve{\imath}\else \u{\i}\fi}'; + $texmacro[0x12E]='{\k{I}}'; + $texmacro[0x12F]='{\k{i}}'; + $texmacro[0x130]='{\ifmmode \dot{I}\else \.{I}\fi}'; + $texmacro[0x131]='{\ifmmode \imath \else \i \fi}'; + $texmacro[0x132]='{IJ}'; + $texmacro[0x133]='{ij}'; + $texmacro[0x134]='{\ifmmode \hat{J}\else \^{J}\fi}'; + $texmacro[0x135]='{\ifmmode \hat{\jmath}\else \^{\j}\fi}'; + $texmacro[0x136]='{\ifmmode \mbox{\c{K}}\else \c{K}\fi}'; + $texmacro[0x137]='{\ifmmode \mbox{\c{k}}\else \c{k}\fi}'; + $texmacro[0x138]='{{\fontencoding{LELA}\selectfont\char91}}'; + $texmacro[0x139]='{\ifmmode \acute{L}\else \\\'{L}\fi}'; + $texmacro[0x13A]='{\ifmmode \acute{l}\else \\\'{l}\fi}'; + $texmacro[0x13B]='{\ifmmode \mbox{\c{L}}\else \c{L}\fi}'; + $texmacro[0x13C]='{\ifmmode \mbox{\c{l}}\else \c{l}\fi}'; + $texmacro[0x13D]='{\ifmmode \check{L}\else \v{L}\fi}'; + $texmacro[0x13E]='{\ifmmode \check{l}\else \v{l}\fi}'; + $texmacro[0x13F]='{{\fontencoding{LELA}\selectfont\char201}}'; + $texmacro[0x140]='{{\fontencoding{LELA}\selectfont\char202}}'; + $texmacro[0x141]='{\L}'; + $texmacro[0x142]='{\l}'; + $texmacro[0x143]='{\ifmmode \acute{N}\else \\\'{N}\fi}'; + $texmacro[0x144]='{\ifmmode \acute{n}\else \\\'{n}\fi}'; + $texmacro[0x145]='{\ifmmode \mbox{\c{N}}\else \c{N}\fi}'; + $texmacro[0x146]='{\ifmmode \mbox{\c{n}}\else \c{n}\fi}'; + $texmacro[0x147]='{\ifmmode \check{N}\else \v{N}\fi}'; + $texmacro[0x148]='{\ifmmode \check{n}\else \v{n}\fi}'; + $texmacro[0x149]='{\'n}'; + $texmacro[0x14A]='{\NG}'; + $texmacro[0x14B]='{\ng}'; + $texmacro[0x14C]='{\ifmmode \bar{O}\else \={O}\fi}'; + $texmacro[0x14D]='{\ifmmode \bar{o}\else \={o}\fi}'; + $texmacro[0x14E]='{\ifmmode \breve{O}\else \u{O}\fi}'; + $texmacro[0x14F]='{\ifmmode \breve{o}\else \u{o}\fi}'; + $texmacro[0x150]='{\ifmmode \mbox{\H{O}}\else \H{O}\fi}'; + $texmacro[0x151]='{\ifmmode \mbox{\H{o}}\else \H{o}\fi}'; + $texmacro[0x152]='{\OE}'; + $texmacro[0x153]='{\oe}'; + $texmacro[0x154]='{\ifmmode \acute{R}\else \\\'{R}\fi}'; + $texmacro[0x155]='{\ifmmode \acute{r}\else \\\'{r}\fi}'; + $texmacro[0x156]='{\ifmmode \mbox{\c{R}}\else \c{R}\fi}'; + $texmacro[0x157]='{\ifmmode \mbox{\c{r}}\else \c{r}\fi}'; + $texmacro[0x158]='{\ifmmode \check{R}\else \v{R}\fi}'; + $texmacro[0x159]='{\ifmmode \check{r}\else \v{r}\fi}'; + $texmacro[0x15A]='{\ifmmode \acute{S}\else \\\'{S}\fi}'; + $texmacro[0x15B]='{\ifmmode \acute{s}\else \\\'{s}\fi}'; + $texmacro[0x15C]='{\ifmmode \hat{S}\else \^{S}\fi}'; + $texmacro[0x15D]='{\ifmmode \hat{s}\else \^{s}\fi}'; + $texmacro[0x15E]='{\ifmmode \mbox{\c{S}}\else \c{S}\fi}'; + $texmacro[0x15F]='{\ifmmode \mbox{\c{s}}\else \c{s}\fi}'; + $texmacro[0x160]='{\ifmmode \check{S}\else \v{S}\fi}'; + $texmacro[0x161]='{\ifmmode \check{s}\else \v{s}\fi}'; + $texmacro[0x162]='{\ifmmode \mbox{\c{T}}\else \c{T}\fi}'; + $texmacro[0x163]='{\ifmmode \mbox{\c{t}}\else \c{t}\fi}'; + $texmacro[0x164]='{\ifmmode \check{T}\else \v{T}\fi}'; + $texmacro[0x165]='{\ifmmode \check{t}\else \v{t}\fi}'; + $texmacro[0x166]='{{\fontencoding{LELA}\selectfont\char47}}'; + $texmacro[0x167]='{{\fontencoding{LELA}\selectfont\char63}}'; + $texmacro[0x168]='{\ifmmode \tilde{U}\else \~{U}\fi}'; + $texmacro[0x169]='{\ifmmode \tilde{u}\else \~{u}\fi}'; + $texmacro[0x16A]='{\ifmmode \bar{U}\else \={U}\fi}'; + $texmacro[0x16B]='{\ifmmode \bar{u}\else \={u}\fi}'; + $texmacro[0x16C]='{\ifmmode \breve{U}\else \u{U}\fi}'; + $texmacro[0x16D]='{\ifmmode \breve{u}\else \u{u}\fi}'; + $texmacro[0x16E]='{\ifmmode \mathring{U}\else \r{U}\fi}'; + $texmacro[0x16F]='{\ifmmode \mathring{u}\else \r{u}\fi}'; + $texmacro[0x170]='{\ifmmode \mbox{\H{U}}\else \H{U}\fi}'; + $texmacro[0x171]='{\ifmmode \mbox{\H{u}}\else \H{u}\fi}'; + $texmacro[0x172]='{\ifmmode \k{U}\else \k{U}\fi}'; + $texmacro[0x173]='{\k{u}}'; + $texmacro[0x174]='{\ifmmode \hat{W}\else \^{W}\fi}'; + $texmacro[0x175]='{\ifmmode \hat{w}\else \^{w}\fi}'; + $texmacro[0x176]='{\ifmmode \hat{Y}\else \^{Y}\fi}'; + $texmacro[0x177]='{\ifmmode \hat{y}\else \^{y}\fi}'; + $texmacro[0x178]='{\ifmmode \ddot{Y}\else \"{Y}\fi}'; + $texmacro[0x179]='{\ifmmode \acute{Z}\else \\\'{Z}\fi}'; + $texmacro[0x17A]='{\ifmmode \acute{z}\else \\\'{z}\fi}'; + $texmacro[0x17B]='{\ifmmode \dot{Z}\else \.{Z}\fi}'; + $texmacro[0x17C]='{\ifmmode \dot{z}\else \.{z}\fi}'; + $texmacro[0x17D]='{\ifmmode \check{Z}\else \v{Z}\fi}'; + $texmacro[0x17E]='{\ifmmode \check{z}\else \v{z}\fi}'; + # 1E00..1EFF; Latin Extended Additional + $texmacro[0x1E00]='{\ifmmode \mbox{\ucirc{A}}\else \ucirc{A}\fi}'; + $texmacro[0x1E01]='{\ifmmode \mbox{\ucirc{a}}\else \ucirc{a}\fi}'; + $texmacro[0x1E02]='{\ifmmode \dot{B}\else \.{B}\fi}'; + $texmacro[0x1E03]='{\ifmmode \dot{b}\else \.{b}\fi}'; + $texmacro[0x1E04]='{\ifmmode \mbox{\d{B}}\else \d{B}\fi}'; + $texmacro[0x1E05]='{\ifmmode \mbox{\d{b}}\else \d{b}\fi}'; + $texmacro[0x1E06]='{\ifmmode \mbox{\b{B}}\else \b{B}\fi}'; + $texmacro[0x1E07]='{\ifmmode \mbox{\b{b}}\else \b{b}\fi}'; + $texmacro[0x1E08]='{\ifmmode \mbox{\\\'{\c{C}}}\else \\\'{\c{C}}\fi}'; + $texmacro[0x1E09]='{\ifmmode \mbox{\\\'{\c{c}}}\else \\\'{\c{c}}\fi}'; + $texmacro[0x1E0A]='{\ifmmode \dot{D}\else \.{D}\fi}'; + $texmacro[0x1E0B]='{\ifmmode \dot{d}\else \.{d}\fi}'; + $texmacro[0x1E0C]='{\ifmmode \mbox{\d{D}}\else \d{D}\fi}'; + $texmacro[0x1E0D]='{\ifmmode \mbox{\d{d}}\else \d{d}\fi}'; + $texmacro[0x1E0E]='{\ifmmode \mbox{\b{D}}\else \b{D}\fi}'; + $texmacro[0x1E0F]='{\ifmmode \mbox{\b{d}}\else \b{d}\fi}'; + $texmacro[0x1E10]='{\ifmmode \mbox{\c{D}}\else \c{D}\fi}'; + $texmacro[0x1E11]='{\ifmmode \mbox{\c{d}}\else \c{d}\fi}'; + $texmacro[0x1E12]='{\ifmmode \mbox{\ucircumflex{D}}\else \ucircumflex{D}\fi}'; + $texmacro[0x1E13]='{\ifmmode \mbox{\ucircumflex{d}}\else \ucircumflex{d}\fi}'; + $texmacro[0x1E14]='{\ifmmode \mbox{\MacGrave{E}}\else \MacGrave{E}\fi}'; + $texmacro[0x1E15]='{\ifmmode \mbox{\MacGrave{e}}\else \MacGrave{e}\fi}'; + $texmacro[0x1E16]='{\ifmmode \mbox{\MacAcute{E}}\else \MacAcute{E}\fi}'; + $texmacro[0x1E17]='{\ifmmode \mbox{\MacAcute{e}}\else \MacAcute{e}\fi}'; + $texmacro[0x1E18]='{\ifmmode \mbox{\ucircumflex{E}}\else \ucircumflex{E}\fi}'; + $texmacro[0x1E19]='{\ifmmode \mbox{\ucircumflex{e}}\else \ucircumflex{e}\fi}'; + $texmacro[0x1E1A]='{\ifmmode \mbox{\utilde{E}}\else \utilde{E}\fi}'; + $texmacro[0x1E1B]='{\ifmmode \mbox{\utilde{e}}\else \utilde{e}\fi}'; + $texmacro[0x1E1C]='{\ifmmode \mbox{\c{\u{E}}}\else \c{\u{E}}\fi}'; + $texmacro[0x1E1D]='{\ifmmode \mbox{\c{\u{e}}}\else \c{\u{e}}\fi}'; + $texmacro[0x1E1E]='{\ifmmode \dot{F}\else \.{F}\fi}'; + $texmacro[0x1E1F]='{\ifmmode \dot{f}\else \.{f}\fi}'; + $texmacro[0x1E20]='{\ifmmode \bar{G}\else \={G}\fi}'; + $texmacro[0x1E21]='{\ifmmode \bar{g}\else \={g}\fi}'; + $texmacro[0x1E22]='{\ifmmode \dot{H}\else \.{H}\fi}'; + $texmacro[0x1E23]='{\ifmmode \dot{h}\else \.{h}\fi}'; + $texmacro[0x1E24]='{\ifmmode \mbox{\d{H}}\else \d{H}\fi}'; + $texmacro[0x1E25]='{\ifmmode \mbox{\d{h}}\else \d{h}\fi}'; + $texmacro[0x1E26]='{\ifmmode \ddot{H}\else \"{H}\fi}'; + $texmacro[0x1E27]='{\ifmmode \ddot{h}\else \"{h}\fi}'; + $texmacro[0x1E28]='{\ifmmode \mbox{\c{H}}\else \c{H}\fi}'; + $texmacro[0x1E29]='{\ifmmode \mbox{\c{h}}\else \c{h}\fi}'; + $texmacro[0x1E2A]='{\ifmmode \mbox{\ubreve{H}}\else \ubreve{H}\fi}'; + $texmacro[0x1E2B]='{\ifmmode \mbox{\ubreve{h}}\else \ubreve{h}\fi}'; + $texmacro[0x1E2C]='{\ifmmode \mbox{\utilde{I}}\else \utilde{I}\fi}'; + $texmacro[0x1E2D]='{\ifmmode \mbox{\utilde{i}}\else \utilde{i}\fi}'; + $texmacro[0x1E2E]='{\ifmmode \mbox{\DiaAcute{I}}\else \DiaAcute{I}\fi}'; + $texmacro[0x1E2F]='{\ifmmode \mbox{\DiaAcute{\i}}\else \DiaAcute{\i}\fi}'; + $texmacro[0x1E30]='{\ifmmode \acute{K}\else \\\'{K}\fi}'; + $texmacro[0x1E31]='{\ifmmode \acute{k}\else \\\'{k}\fi}'; + $texmacro[0x1E32]='{\ifmmode \mbox{\d{K}}\else \d{K}\fi}'; + $texmacro[0x1E33]='{\ifmmode \mbox{\d{k}}\else \d{k}\fi}'; + $texmacro[0x1E34]='{\ifmmode \mbox{\b{K}}\else \b{K}\fi}'; + $texmacro[0x1E35]='{\ifmmode \mbox{\b{k}}\else \b{k}\fi}'; + $texmacro[0x1E36]='{\ifmmode \mbox{\d{L}}\else \d{L}\fi}'; + $texmacro[0x1E37]='{\ifmmode \mbox{\d{l}}\else \d{l}\fi}'; + $texmacro[0x1E38]='{\ifmmode \udot{\bar{L}}\else \udot{\={L}}\fi}'; + $texmacro[0x1E39]='{\ifmmode \udot{\bar{l}}\else \udot{\={l}}\fi}'; + $texmacro[0x1E3A]='{\ifmmode \mbox{\b{L}}\else \b{L}\fi}'; + $texmacro[0x1E3B]='{\ifmmode \mbox{\b{l}}\else \b{l}\fi}'; + $texmacro[0x1E3C]='{\ifmmode \mbox{\ucircumflex{L}}\else \ucircumflex{L}\fi}'; + $texmacro[0x1E3D]='{\ifmmode \mbox{\ucircumflex{l}}\else \ucircumflex{l}\fi}'; + $texmacro[0x1E3E]='{\ifmmode \acute{M}\else \\\'{M}\fi}'; + $texmacro[0x1E3F]='{\ifmmode \acute{m}\else \\\'{m}\fi}'; + $texmacro[0x1E40]='{\ifmmode \dot{M}\else \.{M}\fi}'; + $texmacro[0x1E41]='{\ifmmode \dot{m}\else \.{m}\fi}'; + $texmacro[0x1E42]='{\ifmmode \mbox{\d{M}}\else \d{M}\fi}'; + $texmacro[0x1E43]='{\ifmmode \mbox{\d{m}}\else \d{m}\fi}'; + $texmacro[0x1E44]='{\ifmmode \dot{N}\else \.{N}\fi}'; + $texmacro[0x1E45]='{\ifmmode \dot{n}\else \.{n}\fi}'; + $texmacro[0x1E46]='{\ifmmode \mbox{\d{N}}\else \d{N}\fi}'; + $texmacro[0x1E47]='{\ifmmode \mbox{\d{n}}\else \d{n}\fi}'; + $texmacro[0x1E48]='{\ifmmode \mbox{\b{N}}\else \b{N}\fi}'; + $texmacro[0x1E49]='{\ifmmode \mbox{\b{n}}\else \b{n}\fi}'; + $texmacro[0x1E4A]='{\ifmmode \mbox{\ucircumflex{N}}\else \ucircumflex{N}\fi}'; + $texmacro[0x1E4B]='{\ifmmode \mbox{\ucircumflex{n}}\else \ucircumflex{n}\fi}'; + $texmacro[0x1E4C]='{\ifmmode \mbox{\TilAcute{O}}\else \TilAcute{O}\fi}'; + $texmacro[0x1E4D]='{\ifmmode \mbox{\TilAcute{o}}\else \TilAcute{o}\fi}'; + $texmacro[0x1E4E]='{\ifmmode \mbox{\TilDiaresis{O}}\else \TilDiaresis{O}\fi}'; + $texmacro[0x1E4F]='{\ifmmode \mbox{\TilDiaresis{o}}\else \TilDiaresis{o}\fi}'; + $texmacro[0x1E50]='{\ifmmode \mbox{\MacGrave{O}}\else \MacGrave{O}\fi}'; + $texmacro[0x1E51]='{\ifmmode \mbox{\MacGrave{o}}\else \MacGrave{o}\fi}'; + $texmacro[0x1E52]='{\ifmmode \mbox{\MacAcute{O}}\else \MacAcute{O}\fi}'; + $texmacro[0x1E53]='{\ifmmode \mbox{\MacAcute{o}}\else \MacAcute{o}\fi}'; + $texmacro[0x1E54]='{\ifmmode \acute{P}\else \\\'{P}\fi}'; + $texmacro[0x1E55]='{\ifmmode \acute{p}\else \\\'{p}\fi}'; + $texmacro[0x1E56]='{\ifmmode \dot{P}\else \.{P}\fi}'; + $texmacro[0x1E57]='{\ifmmode \dot{p}\else \.{p}\fi}'; + $texmacro[0x1E58]='{\ifmmode \dot{R}\else \.{R}\fi}'; + $texmacro[0x1E59]='{\ifmmode \dot{r}\else \.{r}\fi}'; + $texmacro[0x1E5A]='{\ifmmode \mbox{\d{R}}\else \d{R}\fi}'; + $texmacro[0x1E5B]='{\ifmmode \mbox{\d{r}}\else \d{r}\fi}'; + $texmacro[0x1E5C]='{\ifmmode \udot{\bar{R}}\else \udot{\={R}}\fi}'; + $texmacro[0x1E5D]='{\ifmmode \udot{\bar{r}}\else \udot{\={r}}\fi}'; + $texmacro[0x1E5E]='{\ifmmode \mbox{\b{R}}\else \b{R}\fi}'; + $texmacro[0x1E5F]='{\ifmmode \mbox{\b{r}}\else \b{r}\fi}'; + $texmacro[0x1E60]='{\ifmmode \dot{S}\else \.{S}\fi}'; + $texmacro[0x1E61]='{\ifmmode \dot{s}\else \.{s}\fi}'; + $texmacro[0x1E62]='{\ifmmode \mbox{\d{S}}\else \d{S}\fi}'; + $texmacro[0x1E63]='{\ifmmode \mbox{\d{s}}\else \d{s}\fi}'; + $texmacro[0x1E64]='{\ifmmode \acute{S}\else \\\'{S}\fi}'; + $texmacro[0x1E65]='{\ifmmode \acute{s}\else \\\'{s}\fi}'; + $texmacro[0x1E66]='{\ifmmode \mbox{\CaronDot{S}}\else \CaronDot{S}\fi}'; + $texmacro[0x1E67]='{\ifmmode \mbox{\CaronDot{s}}\else \CaronDot{s}\fi}'; + $texmacro[0x1E68]='{\ifmmode \udot{\dot{S}}\else \udot{\.{S}}\fi}'; + $texmacro[0x1E69]='{\ifmmode \udot{\dot{s}}\else \udot{\.{s}}\fi}'; + $texmacro[0x1E6A]='{\ifmmode \dot{T}\else \.{T}\fi}'; + $texmacro[0x1E6B]='{\ifmmode \dot{t}\else \.{t}\fi}'; + $texmacro[0x1E6C]='{\ifmmode \mbox{\d{T}}\else \d{T}\fi}'; + $texmacro[0x1E6D]='{\ifmmode \mbox{\d{t}}\else \d{t}\fi}'; + $texmacro[0x1E6E]='{\ifmmode \mbox{\b{T}}\else \b{T}\fi}'; + $texmacro[0x1E6F]='{\ifmmode \mbox{\b{t}}\else \b{t}\fi}'; + $texmacro[0x1E70]='{\ifmmode \mbox{\ucircumflex{T}}\else \ucircumflex{T}\fi}'; + $texmacro[0x1E71]='{\ifmmode \mbox{\ucircumflex{t}}\else \ucircumflex{t}\fi}'; + $texmacro[0x1E72]='{\ifmmode \mbox{\udiaresis{U}}\else \udiaresis{U}\fi}'; + $texmacro[0x1E73]='{\ifmmode \mbox{\udiaresis{u}}\else \udiaresis{u}\fi}'; + $texmacro[0x1E74]='{\ifmmode \mbox{\utilde{U}}\else \utilde{U}\fi}'; + $texmacro[0x1E75]='{\ifmmode \mbox{\utilde{u}}\else \utilde{u}\fi}'; + $texmacro[0x1E76]='{\ifmmode \mbox{\ucircumflex{U}}\else \ucircumflex{U}\fi}'; + $texmacro[0x1E77]='{\ifmmode \mbox{\ucircumflex{u}}\else \ucircumflex{u}\fi}'; + $texmacro[0x1E78]='{\ifmmode \mbox{\TilAcute{U}}\else \TilAcute{U}\fi}'; + $texmacro[0x1E79]='{\ifmmode \mbox{\TilAcute{u}}\else \TilAcute{u}\fi}'; + $texmacro[0x1E7A]='{\ifmmode \mbox{\MacDiaresis{U}}\else \MacDiaresis{U}\fi}'; + $texmacro[0x1E7B]='{\ifmmode \mbox{\MacDiaresis{u}}\else \MacDiaresis{u}\fi}'; + $texmacro[0x1E7C]='{\ifmmode \tilde{V}\else \~{V}\fi}'; + $texmacro[0x1E7D]='{\ifmmode \tilde{v}\else \~{v}\fi}'; + $texmacro[0x1E7E]='{\ifmmode \mbox{\d{V}}\else \d{V}\fi}'; + $texmacro[0x1E7F]='{\ifmmode \mbox{\d{v}}\else \d{v}\fi}'; + $texmacro[0x1E80]='{\ifmmode \grave{W}\else \`{W}\fi}'; + $texmacro[0x1E81]='{\ifmmode \grave{w}\else \`{w}\fi}'; + $texmacro[0x1E82]='{\ifmmode \acute{W}\else \\\'{W}\fi}'; + $texmacro[0x1E83]='{\ifmmode \acute{w}\else \\\'{w}\fi}'; + $texmacro[0x1E84]='{\ifmmode \ddot{W}\else \"{W}\fi}'; + $texmacro[0x1E85]='{\ifmmode \ddot{w}\else \"{w}\fi}'; + $texmacro[0x1E86]='{\ifmmode \dot{W}\else \.{W}\fi}'; + $texmacro[0x1E87]='{\ifmmode \dot{w}\else \.{w}\fi}'; + $texmacro[0x1E88]='{\ifmmode \mbox{\d{W}}\else \d{W}\fi}'; + $texmacro[0x1E89]='{\ifmmode \mbox{\d{w}}\else \d{w}\fi}'; + $texmacro[0x1E8A]='{\ifmmode \dot{X}\else \.{X}\fi}'; + $texmacro[0x1E8B]='{\ifmmode \dot{x}\else \.{x}\fi}'; + $texmacro[0x1E8C]='{\ifmmode \mbox{\d{X}}\else \d{X}\fi}'; + $texmacro[0x1E8D]='{\ifmmode \mbox{\d{x}}\else \d{x}\fi}'; + $texmacro[0x1E8E]='{\ifmmode \dot{Y}\else \.{Y}\fi}'; + $texmacro[0x1E8F]='{\ifmmode \dot{y}\else \.{y}\fi}'; + $texmacro[0x1E90]='{\ifmmode \hat{Z}\else \^{Z}\fi}'; + $texmacro[0x1E91]='{\ifmmode \hat{z}\else \^{z}\fi}'; + $texmacro[0x1E92]='{\ifmmode \mbox{\d{Z}}\else \d{Z}\fi}'; + $texmacro[0x1E93]='{\ifmmode \mbox{\d{z}}\else \d{z}\fi}'; + $texmacro[0x1E94]='{\ifmmode \mbox{\b{Z}}\else \b{Z}\fi}'; + $texmacro[0x1E95]='{\ifmmode \mbox{\b{z}}\else \b{z}\fi}'; + $texmacro[0x1E96]='{\ifmmode \mbox{\b{h}}\else \b{h}\fi}'; + $texmacro[0x1E97]='{\ifmmode \ddot{t}\else \"{t}\fi}'; + $texmacro[0x1E98]='{\ifmmode \mathring{w}\else \r{w}\fi}'; + $texmacro[0x1E99]='{\ifmmode \mathring{y}\else \r{y}\fi}'; +} diff --git a/inCHISE-5.6 b/inCHISE-5.6 new file mode 100755 index 0000000..dd9411d --- /dev/null +++ b/inCHISE-5.6 @@ -0,0 +1,587 @@ +#!/usr/bin/perl + +# ver.0.2 + +use strict; +use vars qw($omegadb_path + $opt_protrude $opt_order + $opt_in_cs $opt_out_cs + $opt_help $usage + $in_cs $out_cs $i @chars + @order %order + $char $char_id $out_char + $ids $ids_argc %ids $idsdb + $idsdata_file $ids_start $font_start + $inotp $perl56 $perl58 + @CDP @HZK @GT + ); +use Getopt::Long; +use utf8; +use Fcntl ':flock'; +use Chise_utils ':all'; + +my $omegadb_path="/usr/local/lib/chise/omega"; + +### Options ### + +$opt_order='jcgk'; +#$opt_order='jtcgkhd'; +$opt_protrude=0;# 1=true, 0=false. + +### +my $strictly_forbidden_after = '「【『［（〈“‘‘（〔｛《{\[\(\x{3016}｛｢'; +# \x{3016} | # white 【 + +my $forbidden_after = "\x{0000}"; + +# ￥¥＄$〒♯＃#￠¢￡£＠@§ +my $slightly_forbidden_after = '￥¥＄$〒♯＃#￠¢￡£＠@§'; + +# $strictly_forbidden_before +# All these characters are allowed to protrude +# in the right margin +my $strictly_forbidden_before= + '!,.:;?、。！，．：；？。\)#}’”〉》」』】〕\x{3017}）］｝｝」\]'; +### \x{3017} | # white 】 + +my $forbidden_before + = 'ー々ぁぃぅぇぉゃゅょっゎァィゥェォャュョッヮヵヶ'; + +my $slightly_forbidden_before + = '\x{000a}\#\-‐−‰′″℃゛゜ゝゞヽヾ＂％－ﾞﾟ'; + +my $asian = '\x{1100}-\x{11FF}\x{2E80}-\x{D7AF}\x{F900}-\x{FAFF}\x{FE30}-\x{FE4F}\x{FF00}-'; + +my $space = '\x{0020}\x{0009}\x{000A}\x{000C}\x{000D}'; + +if($^V and $^V ge v5.8){ + $perl58=1; +}elsif($^V and $^V ge v5.6){ + $perl56=1; +}else{ + print STDERR "This versin is not supported."; + exit 1; +} +if($perl58){ + eval "use Encode"; + binmode(STDIN, ':encoding(utf8)'); + binmode(STDOUT, ':encoding(utf8)'); +} + +&GetOptions("in=s"=>\$opt_in_cs, + "out=s"=>\$opt_out_cs, + "help",\$opt_help); + +$usage=< -o + input coding system: + Utf8mcs, Utf8cns, Utf8gb, Utf8jis, Utf8ks + cmap encoding: + UniCNS, UniGB, UniJIS, UniKS, UniMulti +EOF + +if($opt_in_cs or $opt_out_cs){ + $in_cs=$opt_in_cs; + $out_cs=$opt_out_cs; +}elsif(@ARGV==0){ + ($in_cs,$out_cs)=($0=~/(Utf8.+)To(\w+)/); + $inotp=1; +} + +# $in_cs: +# Utf8mcs,Utf8cns,Utf8gb,Utf8jis,Utf8ks, +# $out_cs: +# UniCNS,UniGB,UniJIS,UniKS,UniMulti + +$in_cs=~s/Utf8/ucs\@/; + +if($opt_help + or not defined($in_cs) + or not defined($out_cs)){ + print $usage; + exit 1; +} + +$omegadb_path=~s!/$!!; + +$idsdata_file="$omegadb_path/idsdata.pl"; +$ids_start=0x00; +$font_start=0; + +if(-e $idsdata_file){ + open(IDSDATA,"+<$idsdata_file") or die; + flock(IDSDATA,LOCK_EX); + seek(IDSDATA,0,0); + while(){ + eval $_; + } + seek(IDSDATA,0,0); +# require $idsdata_file; +}else{ + open(IDSDATA,">$idsdata_file") or die; + flock(IDSDATA,LOCK_EX); + seek(IDSDATA,0,0); +} + +$ids_argc=0; +$ids=""; + +@GT=(#"=gt","=gt-k", + "=gt-pj-1","=gt-pj-2","=gt-pj-3","=gt-pj-4","=gt-pj-5","=gt-pj-6","=gt-pj-7","=gt-pj-8","=gt-pj-9","=gt-pj-10","=gt-pj-11" + #,"=gt-pj-k1","=gt-pj-k2" + ); +@HZK=("=hanziku-1","=hanziku-2","=hanziku-3","=hanziku-4","=hanziku-5","=hanziku-6","=hanziku-7","=hanziku-8","=hanziku-9","=hanziku-10","=hanziku-11","=hanziku-12"); +@CDP=("=big5-cdp"); + +%order=('c'=>'UniCNS', + 'g'=>'UniGB', + 'j'=>'UniJIS', + 'k'=>'UniKS', + 't'=>'GT', + 'h'=>'HZK', + 'd'=>'CDP', + ); + +if(defined($opt_order)){ + if($opt_order=~/^[cgjkthd]*$/){ + @order=split(//,$opt_order); + @order=map {$order{$_}} @order; + }else{ + print STDERR "Invalid order!\n"; + exit 1; + } +} + +while(<>){ + if($perl56){ + # for perl 5.6 + if($inotp){ + s/(.)/pack("c",&get_char_id(unpack("U",$1),$in_cs))/ge; + }else{ + s/(.)/pack("U",&get_char_id(unpack("U",$1),$in_cs))/ge; + } + }elsif($perl58){ + # for perl 5.8. + $_=decode('utf8', $_); + s/(.)/pack("U",&get_char_id(unpack("U",$1),$in_cs))/ge; + } + s/(amp.+?;)/&de_er($1)/ge; +# s/(&.+?;)/&de_er($1)/ge; + @chars=split(//); + for($i=0;$i<=$#chars;$i++){ + $char=$chars[$i]; + $char_id=unpack("U",$char); + + if($char_id<=0x20){ + print $chars[$i]; + next; + }elsif($char_id>0x20 and $char_id<=0x02af){ + # Basic Latin + # Latin-1 Supplement + # Latin Extended-A + # Latin Extended-B + # IPA Extensions + print &latin_parse(); + next; + }elsif($char_id>=0x2ff0 and $char_id<=0x2fff){ + # Ideographic Description Characters + print &ids_parse(); + next; + }else{ + if(($out_char=&get_output_char($char_id,$out_cs))){ + print $out_char,&add_break($i); + }elsif($char_id >= 0x20000 && $char_id <=0x2a6df){ + # CJK Unified Ideographs Extension B + if(not defined($ids{$char}) and $ids{$char}[1]>=0){ + $ids{$char}[0]=$font_start; + $ids{$char}[1]=$ids_start; + $ids_start++; + if($ids_start>255){ + $ids_start=0; + $font_start++; + } + } + print "{\\fontencoding{OT1}\\fontfamily{" . + sprintf("chise%03d",$ids{$char}[0]) . + "}\\selectfont\\char$ids{$char}[1]}",&add_break($i); + next; + }else{ + if($ids=&get_ids($char)){ + print &get_macro_for_ids($ids),&add_break($i); + }else{ + print '\rule{1ex}{1ex}',&add_break($i); + } + } + } + } +} + +print IDSDATA 'use utf8;',"\n"; +foreach $ids (keys %ids){ + print IDSDATA '$ids{\'',$ids,'\'}=[',join ",",@{$ids{$ids}},"];\n" if($perl56); + print IDSDATA '$ids{\'',encode('utf8',$ids),'\'}=[',join ",",@{$ids{$ids}},"];\n" if($perl58); +} +print IDSDATA '$font_start=',$font_start,";\n"; +print IDSDATA '$ids_start=',$ids_start,";\n"; +print IDSDATA "1;"; +flock(IDSDATA,LOCK_UN); + +exit 0; + +sub add_break{ + my($i)=@_; + + if($i<($#chars-1)){ + if(($chars[$i+1]=~m/[$strictly_forbidden_before]/o) + and($chars[$i+2]=~m/[$strictly_forbidden_before]/o)){ + return "\\CJKunbreakablekernone "; + }elsif($opt_protrude){ + if(($chars[$i+1]=~m/[$strictly_forbidden_before]/o) + and($chars[$i+2]=~m/[^$strictly_forbidden_before]/o)){ + return "\\CJKunbreakablekernone \\CJKprotrude "; + } + } + } + if(($i<$#chars) + and($chars[$i+1]=~m/[$strictly_forbidden_before]/o)){ + return "\\CJKunbreakablekernone "; + } + if($chars[$i]=~m/[$strictly_forbidden_after]/o){ + return "\\CJKunbreakablekernone "; + } + if(($i<$#chars) + and($chars[$i+1]=~m/[$forbidden_before]/o)){ + return "\\CJKunbreakablekerntwo "; + + } + if($chars[$i]=~m/[$forbidden_after]/o){ + return "\\CJKunbreakablekerntwo "; + } + if(($i<$#chars) + and($chars[$i+1]=~m/[$slightly_forbidden_before]/o)){ + return "\\CJKunbreakablekernthree "; + } + if($chars[$i]=~m/[$slightly_forbidden_after]/o){ + return "\\CJKunbreakablekernthree "; + } + if($chars[$i]=~m/[$asian]/o){ + return "\\CJKbreakablekern "; + } + if(($i<$#chars)and($chars[$i+1]=~m/[$asian]/o)){ + return "\\CJKbreakablekern "; + } +} + +sub latin_parse{ + # arguments: none + # return: string for output with TeX macro. + my($char_id); + my $out_str=$chars[$i]; + $i++; + while($i<=$#chars){ + $char_id=unpack("U",$chars[$i]); + if($char_id>0x20 and $char_id<=0x02af){ + $out_str.=pack("U",$char_id); + }else{ + $i--; + last; + } + $i++; + } + return '{\normalfont {'.$out_str.'}}'; +} + +sub ids_parse{ + # arguments: none + # return: character for output, + # TeX macro for ids, + # or GETA character if ids is invalid. + my($ids,$ids_argc)=&ids_rest("",0,$chars[$i]); + + while($ids_argc>0){ + # We are in IDS. + $i++; + if($i>$#chars){ + print STDERR "IDS parse error: $ids\n"; +# return pack("U",0xfffd); + return pack("U",0x3013); + } + + ($ids,$ids_argc)=&ids_rest($ids,$ids_argc,$chars[$i]); + if($ids_argc==0){ + if(($char_id=&get_char_id_for_ids($ids)) + and($out_char=&get_output_char($char_id,$out_cs))){ + return $out_char; + }else{ + return &get_macro_for_ids($ids) if($perl56); + return encode('utf8', &get_macro_for_ids($ids)) if($perl58); + } + } + } +} + +sub ids_rest{ + # arguments: , , + # return: ids and rest number of arguments for ids. + my($ids,$ids_argc,$char)=@_; + my($argc); + $argc=&ids_argc($char); + if($argc){ + $ids_argc+= $ids_argc==0 ? $argc : $argc-1; + }else{ + $ids_argc--; + } + $ids.=$char if($perl56); + $ids.=encode('utf8',$char) if($perl58); + return ($ids,$ids_argc); +} + +sub get_macro_for_ids{ + # argument: + # return: TeX macro for ids + # or GETA character if ids is invalid for KAGE. + my($ids)=@_; + $ids=&normalize_ids($ids,"UniJIS"); +# return pack("U",0xfffd) if(($ids!~/[$idc]/) + return pack("U",0x3013) if(($ids!~/[$idc]/) + or($ids=~/[\x{10000}-]/)); + #irregular for KAGE. + if(not defined($ids{$ids}) and $ids{$ids}[1]>=0){ + $ids{$ids}[0]=$font_start; + $ids{$ids}[1]=$ids_start; + $ids_start++; + } + if($ids_start>255){ + $ids_start=0; + $font_start++; + } + return "{\\fontencoding{OT1}\\fontfamily{" + .sprintf("chise%03d",$ids{$ids}[0]) + ."}\\selectfont\\char$ids{$ids}[1]}"; +} + +sub normalize_ids{ + # argument: , + # return: ids or GETA character if ids is invalid for KAGE. + my($ids,$out_cs)=@_; + $ids = decode('utf8', $ids) if $perl58; + $out_cs=~s/Uni(.+)/'ucs@'.lc($1)/e; + + my $output_ids=""; + my($char,$char_id,$output_char_id); + while($ids=~m/(.)/g){ + $char=$1; + $char_id=unpack("U",$char); + if($char=~/[$idc]/){ + $output_ids.=$char; + }elsif($output_char_id=&get_char_attribute($char,"=$out_cs")){ + $output_ids.=pack("U",$output_char_id); + }elsif($output_char_id=&get_char_attribute($char,"=ucs")){ + $output_ids.=pack("U",$output_char_id); + }else{ +# return pack("U",0xfffd); + return pack("U",0x3013); + } + } + return $output_ids; +} + +sub get_output_char{ + # argument: + # return: character in output coding system or TeX macro. + my($char_id,$out_cs)=@_; + my($char,$out_char_id,$suffix); + my($gt,$hzk,$cdp); + + $char=pack('U',$char_id); + + if($out_cs eq 'UniJIS' + and &get_char_attribute($char,"vnd-adobe-cid-unijis-utf16-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@jis') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@jis') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return pack("U",$out_char_id); + } + }elsif($out_cs eq 'UniGB' + and &get_char_attribute($char,"vnd-adobe-cid-unigb-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@gb') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@gb') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return pack("U",$out_char_id); + } + }elsif($out_cs eq 'UniCNS' + and &get_char_attribute($char,"vnd-adobe-cid-unicns-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@cns') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@cns') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return pack("U",$out_char_id); + } + }elsif($out_cs eq 'UniKS' + and &get_char_attribute($char,"vnd-adobe-cid-uniks-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@ks') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@ks') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return pack("U",$out_char_id); + } + }elsif($out_cs eq 'UniMulti'){ + foreach $out_cs (@order){ + + if($out_cs eq 'UniJIS' + and &get_char_attribute($char,"vnd-adobe-cid-unijis-utf16-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@jis') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@jis') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return '{\selectjisfont\char'.$out_char_id.'}'; + } + }elsif($out_cs eq 'UniGB' + and &get_char_attribute($char,"vnd-adobe-cid-unigb-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@gb') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@gb') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return '{\selectgbsfont\char'.$out_char_id.'}'; + } + }elsif($out_cs eq 'UniCNS' + and &get_char_attribute($char,"vnd-adobe-cid-unicns-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@cns') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@cns') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return '{\selectcnsfont\char'.$out_char_id.'}'; + } + }elsif($out_cs eq 'UniKS' + and &get_char_attribute($char,"vnd-adobe-cid-uniks-ucs2-h")){ + if($out_char_id=&get_char_attribute($char,'=ucs@ks') + or $out_char_id=&get_char_attribute($char,'=ucs') + or $out_char_id=&get_char_attribute($char,'=>ucs@ks') + or $out_char_id=&get_char_attribute($char,'=>ucs') + ){ + return '{\selectksxfont\char'.$out_char_id.'}'; + } + }elsif($out_cs eq 'GT'){ + return $gt if($gt=&get_macro_for_GT($char_id)); + }elsif($out_cs eq 'HZK'){ + return $hzk if($hzk=&get_macro_for_HZK($char_id)); + }elsif($out_cs eq 'CDP'){ + return $cdp if($cdp=&get_macro_for_CDP($char_id)); + } + } + } + return undef; +} + +sub get_ids{ + # argument: + # return: ids + my($char)=@_; + my $ids=""; + $ids=&get_char_attribute($char,"ids-aggregated") + or $ids=&get_char_attribute($char,"ids"); +# or $ids=&get_char_attribute($char,"ideographic-structure"); + $ids=decode('utf8', $ids) if($perl58); +# $ids=~s/[? ()]//g; + return $ids; +} + +sub get_char_id_for_ids{ + # argument: + # return: char-id + my($ids)=@_; + my($output_char); + $ids=decode('utf8', $ids) if($perl58); + + if(($output_char)=&get_chars_matching("ids",$ids)){ + return unpack("U",$output_char); + }else{ + return undef; + } +} + +sub get_char_id{ + # argument: , + # return: char-id. + my($char_id,$in_cs)=@_; + my($output_char); + + return $char_id if($in_cs eq 'ucs@mcs'); + + if(($output_char)=&get_chars_matching("=$in_cs",$char_id)){ + return unpack("U",$output_char); + }else{ + return $char_id; + } +} + +sub get_macro_for_GT{ + # argument: + # return: TeX macro for GT fonts. + my($char_id)=@_; + my($char,$gt,$GT); + $char=pack("U",$char_id); + foreach (@GT){ + if($gt=&get_char_attribute($char,$_)){ + m/gt\-pj\-(\d+)/ and $GT=$1; + last; + } + } + if($gt){ + return "{\\fontencoding{OT1}\\fontfamily{" + .sprintf("gt%02d",$GT) + ."}\\selectfont\\char".($gt|0x8080)."}"; + }else{ + return undef; + } +} + +sub get_macro_for_HZK{ + my($char_id)=@_; + my($char,$hzk,$HZK); + $char=pack("U",$char_id); + foreach (@HZK){ + if($hzk=&get_char_attribute($char,$_)){ + m/hanziku\-(\d+)/ and $HZK=$1; + last; + } + } + if($hzk){ + return "{\\fontencoding{OT1}\\fontfamily{".sprintf("hzk%02d",$HZK)."}\\selectfont\\char".$hzk."}"; + }else{ + return undef; + } +} + +sub get_macro_for_CDP{ + my($char_id)=@_; + my($char,$cdp,$ucs); + $char=pack("U",$char_id); + foreach (@CDP){ + if($cdp=&get_char_attribute($char,$_)){ + last; + } + } + if($cdp){ + $ucs=&get_char_attribute(&get_chars_matching("=big5-pua",$cdp),"=ucs"); + if($ucs){ + return "{\\fontencoding{OT1}\\fontfamily{cdp}\\selectfont\\char" + .$ucs. + "}"; + }else{ + print STDERR "This hould not happen.\n"; + print STDERR "ucs code point of CDP: $cdp not found.\n"; + } + }else{ + return undef; + } +}