unicode/ksx1001.pl

   1 #! /usr/bin/perl
   2
   3 # USAGE: perl ksx1001.pl > ksx1001.h
   4
   5 # Requires CP949.TXT, found on:
   6 #   http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP949.TXT"
   7
   8 require "cjkcompat.pl";
   9
  10 $cp949 = 'CP949.TXT';
  11 $perline = 8;
  12
  13 die "${cp949}: File not found.\n" if (!(-f $cp949));
  14 open (SET, $cp949) or die "${cp949}: $!\n";
  15
  16 while (<SET>) {
  17         chomp;
  18
  19         s/\#.*//;
  20
  21         next unless /^0x([0-9A-F]{2,4})\s+0x([0-9A-F]{4})\s*$/;
  22
  23         my ($code, $ucode) = (hex($1), hex($2));
  24
  25         if ($code > 0x8000) {
  26                 print STDERR "Warning: duplicated: $code->$k2u{$code},$ucode\n"
  27                         if defined $k2u{$code};
  28                 $k2u{$code} = $ucode;
  29         }
  30 }
  31
  32 close SET;
  33
  34 # make reversal map.
  35
  36 foreach (keys %k2u) {
  37         print STDERR "Warning: duplicated: $u2k{$k2u{$_}},$_<-$k2u{$_}\n"
  38                 if defined $u2k{$k2u{$_}};
  39         $c1 = $_ >> 8;
  40         $c2 = $_ & 0x00FF;
  41         if ($c1 >= 0xA1 && $c2 >= 0xA1) {
  42                 $u2k_ksx1001{$k2u{$_}} = $_;
  43         } else {
  44                 $u2k_cp949{$k2u{$_}} = $_;
  45         }
  46 }
  47
  48 # Add maps for CJK compatibility ideographs of Unicode.
  49 ####&add_cjkcompat(%compat_ksx1001);
  50 &add_cjkcompat(%compat_big5);
  51 &add_cjkcompat(%compat_ibm32);
  52 &add_cjkcompat(%compat_jisx0213);
  53 &add_cjkcompat(%compat_cns11643);
  54
  55 sub add_cjkcompat {
  56 local(%compat) = @_;
  57   foreach (keys %compat) {
  58     if (defined $u2k_ksx1001{$compat{$_}}) {
  59       $u2k_ksx1001{$_} = $u2k_ksx1001{$compat{$_}};
  60     }
  61     if (defined $u2k_cp949{$compat{$_}}) {
  62       $u2k_cp949{$_} = $u2k_cp949{$compat{$_}};
  63     }
  64   }
  65 }
  66
  67
  68 print <<"EOF";
  69 #ifndef _KSX1001_HDR_
  70 #define _KSX1001_HDR_
  71 /*
  72  * KS X 1001 and CP949 (UHC) support
  73  *  by Hatuka*nezumi - IKEDA Soji <nezumi\@jca.apc.org>
  74  * $Id: ksx1001.pl,v 1.1 2004/02/03 02:00:00 mrsam Exp $
  75  *
  76  */
  77
  78 #include "unicode.h"
  79
  80 #define KS_CHAR_SO      0x0E
  81 #define KS_CHAR_SI      0x0F
  82 #define KS_CHAR_ESC     0x1B
  83
  84 /* ISOREG #1/#3: US-ASCII (identical to ISO 646 IRV) */
  85 #define KS_STATE_ASCII          0x0
  86 /* ISOREG #149: KS X 1001:1992 Wansung */
  87 #define KS_STATE_KSX1001        0x4
  88 /* Unknown state */
  89 #define KS_STATE_BINARY         0xF
  90
  91 EOF
  92
  93 print "/* map: CP949 to Unicode */\n";
  94 for ($hb = 0x81; $hb <= 0xFE; $hb++) {
  95         $items = 0;
  96         for ($lb = 0x41; $lb <= 0xFE; $lb++) {
  97                 $items++ if $k2u{$hb*256 + $lb};
  98         }
  99         if ($items) {
 100                 $items = 0;
 101                 printf "static const unicode_char cp949_to_uni_tbl_%02x[] = {", $hb;
 102                 for ($lb = 0x41; $lb <= 0xFE; $lb++) {
 103                         $code = $hb*256 + $lb;
 104                         print ", " if ($items > 0);
 105                         print "\n  " if ($items % $perline == 0);
 106                         $k2u{$code} = 0xFFFD unless $k2u{$code};
 107                         printf("0x%04X", $k2u{$code});
 108                         $items++;
 109                 }
 110                 print "\n};\n";
 111                 $k2uout{$hb} = 1;
 112         }
 113 }
 114
 115 print "const unicode_char * cp949_to_uni_tbls[] = {\n";
 116 for ($hb=0x81; $hb <= 0xFE; $hb++) {
 117         print (($hb > 0x81) ? ",\n  " : "  ");
 118         if ($k2uout{$hb}) {
 119                 printf "cp949_to_uni_tbl_%02x", $hb;
 120         } else {
 121                 print "NULL";
 122         }
 123 }
 124
 125 print "\n};\n";
 126 print "\n\n";
 127
 128
 129 %u2kout = ();
 130 print "/* map: Unicode to KS X 1001 */\n";
 131 for ($hb = 0x00; $hb <= 0xFF; $hb++) {
 132         $items = 0;
 133         for ($lb = 0x00; $lb <= 0xFF; $lb++) {
 134                 $items++ if $u2k_ksx1001{$hb*256 + $lb};
 135         }
 136         if ($items) {
 137                 $items = 0;
 138                 printf "static const unicode_char uni_to_ksx1001_tbl_%02x[] = {", $hb;
 139                 for ($lb = 0x00; $lb <= 0xFF; $lb++) {
 140                         $code = $hb*256 + $lb;
 141                         print ", " if ($items > 0);
 142                         print "\n  " if ($items % $perline == 0);
 143                         $u2k_ksx1001{$code} = 0x003F unless $u2k_ksx1001{$code};
 144                         printf("0x%04X", $u2k_ksx1001{$code});
 145                         $items++;
 146                 }
 147                 print "\n};\n";
 148                 $u2kout{$hb} = 1;
 149         }
 150 }
 151
 152 print "const unicode_char * uni_to_ksx1001_tbls[] = {\n";
 153 for ($hb=0x00; $hb <= 0xFF; $hb++) {
 154         print (($hb > 0x00) ? ",\n  " : "  ");
 155         if ($u2kout{$hb}) {
 156                 printf "uni_to_ksx1001_tbl_%02x", $hb;
 157         } else {
 158                 print "NULL";
 159         }
 160 }
 161
 162 print "\n};\n";
 163 print "\n\n";
 164
 165 %u2kout = ();
 166 print "/* map: Unicode to CP949 extension */\n";
 167 for ($hb = 0x00; $hb <= 0xFF; $hb++) {
 168         $items = 0;
 169         for ($lb = 0x00; $lb <= 0xFF; $lb++) {
 170                 $items++ if $u2k_cp949{$hb*256 + $lb};
 171         }
 172         if ($items) {
 173                 $items = 0;
 174                 printf "static const unicode_char uni_to_cp949_tbl_%02x[] = {", $hb;
 175                 for ($lb = 0x00; $lb <= 0xFF; $lb++) {
 176                         $code = $hb*256 + $lb;
 177                         print ", " if ($items > 0);
 178                         print "\n  " if ($items % $perline == 0);
 179                         $u2k_cp949{$code} = 0x003F unless $u2k_cp949{$code};
 180                         printf("0x%04X", $u2k_cp949{$code});
 181                         $items++;
 182                 }
 183                 print "\n};\n";
 184                 $u2kout{$hb} = 1;
 185         }
 186 }
 187
 188 print "const unicode_char * uni_to_cp949_tbls[] = {\n";
 189 for ($hb=0x00; $hb <= 0xFF; $hb++) {
 190         print (($hb > 0x00) ? ",\n  " : "  ");
 191         if ($u2kout{$hb}) {
 192                 printf "uni_to_cp949_tbl_%02x", $hb;
 193         } else {
 194                 print "NULL";
 195         }
 196 }
 197
 198 print "\n};\n";
 199 print "\n\n";
 200
 201
 202 print "#endif /* _KSX1001_HDR_ */\n";