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 open(PRECOMPOSETEMP, ">precompose.TEMP");
  16 open( DECOMPOSETEMP, ">decompose.TEMP");
  17
  18 while (<UNICODEDATA>){
  19     chop;
  20     (
  21      $code0,
  22      $Name1,
  23      $General_Category2,
  24      $Canonical_Combining_Class3,
  25      $Bidi_Class4,
  26      $Decomposition_Mapping5,
  27      $Numeric_Value6,
  28      $Numeric_Value7,
  29      $Numeric_Value8,
  30      $Bidi_Mirrored9,
  31      $Unicode_1_Name10,
  32      $ISO_Comment11,
  33      $Simple_Uppercase_Mapping12,
  34      $Simple_Lowercase_Mapping13,
  35      $Simple_Titlecase_Mapping14
  36      ) = split(/\;/);
  37
  38     if (($Decomposition_Mapping5 ne "") && ($Decomposition_Mapping5 !~ /\</) && ($Decomposition_Mapping5 =~ / /)) {
  39         ($base, $comb) = split(/ /,$Decomposition_Mapping5);
  40
  41         $leftbracket  = "  { ";
  42         $rightbracket =" },     ";
  43
  44         if (hex($code0) > 0xFFFF) {           # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  45             $leftbracket  = "\/\*{ ";         # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  46             $rightbracket =" },\*\/   ";      # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  47         }                                     # DELETE THIS LINE  IF INTERNAL CODE IS UCS4
  48
  49         printf(PRECOMPOSETEMP "%s0x%08X, 0x%08X, 0x%08X%s\/\* %s \*\/\n", $leftbracket, hex($code0), hex($base), hex($comb), $rightbracket, $Name1);
  50
  51         # AFP 3.x Spec
  52         if ( ((0x2000  <= hex($code0)) && (hex($code0) <=  0x2FFF))
  53           || ((0xFE30  <= hex($code0)) && (hex($code0) <=  0xFE4F))
  54           || ((0x2F800 <= hex($code0)) && (hex($code0) <= 0x2FA1F))) {
  55             $leftbracket  = "\/\*{ ";
  56             $rightbracket =" },\*\/   ";
  57         }
  58
  59         printf( DECOMPOSETEMP "%s0x%08X, 0x%08X, 0x%08X%s\/\* %s \*\/\n", $leftbracket, hex($code0), hex($base), hex($comb), $rightbracket, $Name1);
  60
  61     }
  62 }
  63
  64 # sort ---------------------------------------------------------------------
  65
  66 system("sort -k 3 precompose.TEMP \> precompose.SORT");
  67 system("sort -k 2  decompose.TEMP \>  decompose.SORT");
  68
  69 # print  -------------------------------------------------------------------
  70
  71 printf ("\/\* This file is generated from contrib/misc/make-precompose.h.pl %s \*\/\n", $ARGV[0]);
  72 print ("\/\* DO NOT EDIT BY HAND\!\!\!                                           \*\/\n");
  73 print ("\n");
  74 printf ("\/\* %s is got from                                      \*\/\n", $ARGV[0]);
  75 print ("\/\* http\:\/\/www.unicode.org\/Public\/UNIDATA\/UnicodeData.txt            \*\/\n");
  76 print ("\n");
  77
  78 print ("static const struct \{\n");
  79 print ("  unsigned int replacement\;\n");
  80 print ("  unsigned int base\;\n");
  81 print ("  unsigned int comb\;\n");
  82 print ("\} precompositions\[\] \= \{\n");
  83
  84 system("cat precompose.SORT");
  85
  86 print ("\}\;\n");
  87 print ("\n");
  88
  89 print ("static const struct \{\n");
  90 print ("  unsigned int replacement\;\n");
  91 print ("  unsigned int base\;\n");
  92 print ("  unsigned int comb\;\n");
  93 print ("\} decompositions\[\] \= \{\n");
  94
  95 system("cat decompose.SORT");
  96
  97 print ("\}\;\n");
  98 print ("\n");
  99
 100 print ("\/\* EOF \*\/\n");
 101
 102 # EOF