mcs/class/I18N/CJK/gb2312-build.sh

   1 #!/bin/sh
   2
   3 # Usage: gb2312-build.sh GB2312.TXT > gb2312.table
   4 #
   5 # Get the input file from ftp://ftp.unicode.org/Public/MAPPINGS/OBSOLETE/EASTASIA/GB/GB2312.TXT
   6 #
   7 # This algorithm was taken from the glibc iconv documentation in
   8 # iconvdata/gb2312.c
   9
  10
  11 # GB2312 to Unicode
  12
  13 egrep '^0x' $1 | perl -e \
  14 '
  15 @vals;
  16 while (<>) {
  17     local($gb, $uni, %rest) = split;
  18     local($u)=hex($uni);
  19     local($g)=hex($gb);
  20     #printf("Setting vals[%d] to 0x%04x\n", int(($g - 0x2121) / 256) * 94 + (($g - 0x2121) & 0xff), $u);
  21     @vals[int(($g - 0x2121) / 256) * 94 + (($g - 0x2121) & 0xff)]=$u;
  22 }
  23 $size=($#vals+1)*2;
  24 printf("\001\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
  25 for ($i=0; $i < $#vals+1; $i++) {
  26     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
  27 }
  28 '
  29
  30 # Gb2312 from Unicode, table 1
  31
  32 perl -e \
  33 '
  34 @vals;
  35 @vals[0x00]=0x21 + (0x68 << 8);
  36 @vals[0x03]=0x21 + (0x6c << 8);
  37 @vals[0x04]=0x21 + (0x27 << 8);
  38 @vals[0x0c]=0x21 + (0x63 << 8);
  39 @vals[0x0d]=0x21 + (0x40 << 8);
  40 @vals[0x33]=0x21 + (0x41 << 8);
  41 @vals[0x3c]=0x28 + (0x24 << 8);
  42 @vals[0x3d]=0x28 + (0x22 << 8);
  43 @vals[0x44]=0x28 + (0x28 << 8);
  44 @vals[0x45]=0x28 + (0x26 << 8);
  45 @vals[0x46]=0x28 + (0x3a << 8);
  46 @vals[0x48]=0x28 + (0x2c << 8);
  47 @vals[0x49]=0x28 + (0x2a << 8);
  48 @vals[0x4e]=0x28 + (0x30 << 8);
  49 @vals[0x4f]=0x28 + (0x2e << 8);
  50 @vals[0x53]=0x21 + (0x42 << 8);
  51 @vals[0x55]=0x28 + (0x34 << 8);
  52 @vals[0x56]=0x28 + (0x32 << 8);
  53 @vals[0x58]=0x28 + (0x39 << 8);
  54 @vals[0x5d]=0x28 + (0x21 << 8);
  55
  56 $size=($#vals+1)*2;
  57 printf("\002\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
  58 for ($i=0; $i < $#vals+1; $i++) {
  59     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
  60 }
  61 '
  62
  63 # Gb2312 from Unicode, table 2
  64
  65 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0x03' | perl -e \
  66 '
  67 @vals;
  68 while(<>) {
  69     local($uni, $gb, %rest) = split;
  70     local($u)=hex($uni);
  71     local($g)=hex($gb);
  72     #printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0x391, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
  73     @vals[$u - 0x391]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
  74 }
  75 $size=($#vals+1)*2;
  76 printf("\003\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
  77 for ($i=0; $i < $#vals+1; $i++) {
  78     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
  79 }
  80 '
  81
  82 # Gb2312 from Unicode, table 3
  83
  84 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0x04' | perl -e \
  85 '
  86 @vals;
  87 while(<>) {
  88     local($uni, $gb, %rest) = split;
  89     local($u)=hex($uni);
  90     local($g)=hex($gb);
  91     #printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0x401, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
  92     @vals[$u - 0x401]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
  93 }
  94 $size=($#vals+1)*2;
  95 printf("\004\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
  96 for ($i=0; $i < $#vals+1; $i++) {
  97     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
  98 }
  99 '
 100
 101 # Gb2312 from Unicode, table 4
 102
 103 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0x20' | perl -e \
 104 '
 105 @vals;
 106 while(<>) {
 107     local($uni, $gb, %rest) = split;
 108     local($u)=hex($uni);
 109     local($g)=hex($gb);
 110     #printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0x2015, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
 111     @vals[$u - 0x2015]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
 112 }
 113 $size=($#vals+1)*2;
 114 printf("\005\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
 115 for ($i=0; $i < $#vals+1; $i++) {
 116     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
 117 }
 118 '
 119
 120 # Gb2312 from Unicode, table 5
 121
 122 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0x2[12]' | perl -e \
 123 '
 124 @vals;
 125 while(<>) {
 126     local($uni, $gb, %rest) = split;
 127     local($u)=hex($uni);
 128     local($g)=hex($gb);
 129     #printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0x2103, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
 130     @vals[$u - 0x2103]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
 131 }
 132 $size=($#vals+1)*2;
 133 printf("\006\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
 134 for ($i=0; $i < $#vals+1; $i++) {
 135     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
 136 }
 137 '
 138
 139
 140 # Gb2312 from Unicode, table 6
 141
 142 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0x24' | perl -e \
 143 '
 144 @vals;
 145 while(<>) {
 146     local($uni, $gb, %rest) = split;
 147     local($u)=hex($uni);
 148     local($g)=hex($gb);
 149     #printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0x2460, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
 150     @vals[$u - 0x2460]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
 151 }
 152 $size=($#vals+1)*2;
 153 printf("\007\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
 154 for ($i=0; $i < $#vals+1; $i++) {
 155     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
 156 }
 157 '
 158
 159
 160 # Gb2312 from Unicode, table 7
 161
 162 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0x3[01]' | perl -e \
 163 '
 164 @vals;
 165 while(<>) {
 166     local($uni, $gb, %rest) = split;
 167     local($u)=hex($uni);
 168     local($g)=hex($gb);
 169     #printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0x3000, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
 170     @vals[$u - 0x3000]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
 171 }
 172 $size=($#vals+1)*2;
 173 printf("\010\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
 174 for ($i=0; $i < $#vals+1; $i++) {
 175     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
 176 }
 177 '
 178
 179 # Gb2312 from Unicode, table 8
 180
 181 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0x[4-9]' | perl -e \
 182 '
 183 @vals;
 184 while(<>) {
 185     local($uni, $gb, %rest) = split;
 186     local($u)=hex($uni);
 187     local($g)=hex($gb);
 188     printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0x4e00, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
 189     @vals[$u - 0x4e00]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
 190 }
 191 $size=($#vals+1)*2;
 192 printf("\011\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
 193 for ($i=0; $i < $#vals+1; $i++) {
 194     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
 195 }
 196 '
 197
 198 # Gb2312 from Unicode, table 9
 199
 200 egrep '^0x' $1 | awk '{ print $2, $1 }' | sort | egrep '^0xFF[0-5]' | perl -e \
 201 '
 202 @vals;
 203 while(<>) {
 204     local($uni, $gb, %rest) = split;
 205     local($u)=hex($uni);
 206     local($g)=hex($gb);
 207     #printf STDERR ("Setting vals[0x%04x] to 0x%04x\n", $u - 0xff01, ($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8));
 208     @vals[$u - 0xff01]=($g < 0x100 ? $g : int($g/256)) + (($g < 0x100 ? 0 : $g&255) << 8);
 209 }
 210 $size=($#vals+1)*2;
 211 printf("\012\000\000\000%c%c%c%c", $size & 0xFF, ($size >> 8) & 0xFF, ($size >> 16) & 0xFF, ($size >> 24) & 0xFF);
 212 for ($i=0; $i < $#vals+1; $i++) {
 213     printf("%c%c", $vals[$i] & 0xFF, ($vals[$i] >> 8) & 0xFF);
 214 }
 215 '
 216
 217