contrib/misc/make-precompose.h.pl

   1 #!/usr/bin/perl
   2
   3 # usage: make-precompose.h.pl UnicodeData.txt > precompose.h
   4
   5 # See
   6 # http://www.unicode.org/Public/UNIDATA/UCD.html
   7 # http://www.unicode.org/reports/tr15/
   8 # http://www.unicode.org/Public/*/ucd/UnicodeData*.txt
   9 # http://www.unicode.org/Public/UNIDATA/UnicodeData.txt
  10
  11
  12 # table for binary search --------------------------------------------------
  13
  14 open(UNICODEDATA, "<$ARGV[0]");
  15
  16 open(PRECOMPOSE_TEMP, ">precompose.TEMP");
  17 open( DECOMPOSE_TEMP, ">decompose.TEMP");
  18
  19 open(PRECOMPOSE_SP_TEMP, ">precompose_sp.TEMP");
  20 open( DECOMPOSE_SP_TEMP, ">decompose_sp.TEMP");
  21
  22 while (<UNICODEDATA>){
  23     chop;
  24     (
  25      $code0,
  26      $Name1,
  27      $General_Category2,
  28      $Canonical_Combining_Class3,
  29      $Bidi_Class4,
  30      $Decomposition_Mapping5,
  31      $Numeric_Value6,
  32      $Numeric_Value7,
  33      $Numeric_Value8,
  34      $Bidi_Mirrored9,
  35      $Unicode_1_Name10,
  36      $ISO_Comment11,
  37      $Simple_Uppercase_Mapping12,
  38      $Simple_Lowercase_Mapping13,
  39      $Simple_Titlecase_Mapping14
  40      ) = split(/\;/);
  41
  42     if (($Decomposition_Mapping5 ne "") && ($Decomposition_Mapping5 !~ /\</) && ($Decomposition_Mapping5 =~ / /)) {
  43         ($base, $comb) = split(/ /,$Decomposition_Mapping5);
  44
  45         $leftbracket  = "  { ";
  46         $rightbracket =" },     ";
  47
  48         # AFP 3.x Spec
  49         if ( ((0x2000  <= hex($code0)) && (hex($code0) <=  0x2FFF))
  50           || ((0xFE30  <= hex($code0)) && (hex($code0) <=  0xFE4F))
  51           || ((0x2F800 <= hex($code0)) && (hex($code0) <= 0x2FA1F))) {
  52             $leftbracket  = "\/\*{ ";
  53             $rightbracket =" },\*\/   ";
  54         }
  55
  56         if (hex($code0) > 0xFFFF) {                            # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  57
  58             $code0_sp_hi = 0xD800 - (0x10000 >> 10) + (hex($code0) >> 10);
  59             $code0_sp_lo = 0xDC00 + (hex($code0) & 0x3FF);
  60
  61             $base_sp_hi = 0xD800 - (0x10000 >> 10) + (hex($base) >> 10);
  62             $base_sp_lo = 0xDC00 + (hex($base) & 0x3FF);
  63
  64             $comb_sp_hi = 0xD800 - (0x10000 >> 10) + (hex($comb) >> 10);
  65             $comb_sp_lo = 0xDC00 + (hex($comb) & 0x3FF);
  66
  67             printf(PRECOMPOSE_SP_TEMP "%s0x%04X%04X, 0x%04X%04X, 0x%04X%04X%s\/\* %s \*\/\n",
  68                    $leftbracket, $code0_sp_hi ,$code0_sp_lo, $base_sp_hi, $base_sp_lo, $comb_sp_hi, $comb_sp_lo, $rightbracket, $Name1);
  69             printf(DECOMPOSE_SP_TEMP "%s0x%04X%04X, 0x%04X%04X, 0x%04X%04X%s\/\* %s \*\/\n",
  70                    $leftbracket, $code0_sp_hi ,$code0_sp_lo, $base_sp_hi, $base_sp_lo, $comb_sp_hi, $comb_sp_lo, $rightbracket, $Name1);
  71
  72             $leftbracket  = "\/\*{ ";                          # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  73             $rightbracket =" },\*\/   ";                       # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  74         }                                                      # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  75
  76         printf(PRECOMPOSE_TEMP "%s0x%08X, 0x%08X, 0x%08X%s\/\* %s \*\/\n", $leftbracket, hex($code0), hex($base), hex($comb), $rightbracket, $Name1);
  77         printf( DECOMPOSE_TEMP "%s0x%08X, 0x%08X, 0x%08X%s\/\* %s \*\/\n", $leftbracket, hex($code0), hex($base), hex($comb), $rightbracket, $Name1);
  78
  79     }
  80 }
  81
  82 # sort ---------------------------------------------------------------------
  83
  84 system("sort -k 3 precompose.TEMP \> precompose.SORT");
  85 system("sort -k 2  decompose.TEMP \>  decompose.SORT");
  86
  87 system("sort -k 3 precompose_sp.TEMP \> precompose_sp.SORT");
  88 system("sort -k 2  decompose_sp.TEMP \>  decompose_sp.SORT");
  89
  90 # print  -------------------------------------------------------------------
  91
  92 printf ("\/\* This file is generated by contrib/misc/make-precompose.h.pl %s \*\/\n", $ARGV[0]);
  93 print ("\/\* DO NOT EDIT BY HAND\!\!\!                                           \*\/\n");
  94 print ("\n");
  95 printf ("\/\* %s is got from                                      \*\/\n", $ARGV[0]);
  96 print ("\/\* http\:\/\/www.unicode.org\/Public\/UNIDATA\/UnicodeData.txt            \*\/\n");
  97 print ("\n");
  98
  99 print ("static const struct \{\n");
 100 print ("  unsigned int replacement\;\n");
 101 print ("  unsigned int base\;\n");
 102 print ("  unsigned int comb\;\n");
 103 print ("\} precompositions\[\] \= \{\n");
 104
 105 system("cat precompose.SORT");
 106
 107 print ("\}\;\n");
 108 print ("\n");
 109
 110 print ("static const struct \{\n");
 111 print ("  unsigned int replacement\;\n");
 112 print ("  unsigned int base\;\n");
 113 print ("  unsigned int comb\;\n");
 114 print ("\} decompositions\[\] \= \{\n");
 115
 116 system("cat decompose.SORT");
 117
 118 print ("\}\;\n");
 119 print ("\n");
 120
 121
 122
 123 print ("static const struct \{\n");
 124 print ("  unsigned int replacement\;\n");
 125 print ("  unsigned int base\;\n");
 126 print ("  unsigned int comb\;\n");
 127 print ("\} precompositions_sp\[\] \= \{\n");
 128
 129 system("cat precompose_sp.SORT");
 130
 131 print ("\}\;\n");
 132 print ("\n");
 133
 134 print ("static const struct \{\n");
 135 print ("  unsigned int replacement\;\n");
 136 print ("  unsigned int base\;\n");
 137 print ("  unsigned int comb\;\n");
 138 print ("\} decompositions_sp\[\] \= \{\n");
 139
 140 system("cat decompose_sp.SORT");
 141
 142 print ("\}\;\n");
 143 print ("\n");
 144
 145 print ("\/\* EOF \*\/\n");
 146
 147 # EOF