unicode/mkultcase.pl

   1 # Copyright 2000-2004 Double Precision, Inc.
   2 # See COPYING for distribution information.
   3 #
   4 # $Id: mkultcase.pl,v 1.3 2004/02/01 06:09:09 mrsam Exp $
   5 #
   6 # Generate unicode upper/lower/titlecase translations.
   7
   8 print '/*
   9 ** Copyright 2000-2004 Double Precision, Inc.
  10 ** See COPYING for distribution information.
  11 **
  12 ** $Id: mkultcase.pl,v 1.3 2004/02/01 06:09:09 mrsam Exp $
  13 */
  14
  15 #include "unicode.h"
  16
  17 ';
  18
  19 my $set=shift;
  20
  21 open (U, "UnicodeData.txt") || die "$!\n";
  22
  23 while (<U>)
  24 {
  25         chomp;
  26
  27 my @fields= split /;/;
  28
  29 my ($code, $uc, $lc, $tc);
  30
  31         $code="0x$fields[0]";
  32         eval "\$code=$code;";
  33
  34         $uc=$fields[12];
  35         if ($uc ne "")
  36         {
  37                 eval "\$uc=0x$uc;";
  38                 $UC{$code}=$uc;
  39                 $FLAG{$code}=1;
  40         }
  41
  42         $lc=$fields[13];
  43         if ($lc ne "")
  44         {
  45                 eval "\$lc=0x$lc;";
  46                 $LC{$code}=$lc;
  47                 $FLAG{$code}=1;
  48         }
  49
  50         $tc=$fields[14];
  51         if ($tc ne "")
  52         {
  53                 eval "\$tc=0x$tc;";
  54                 $TC{$code}=$tc;
  55                 $FLAG{$code}=1;
  56         }
  57 }
  58
  59 close(U);
  60
  61 my $tabsize=1024;
  62
  63 grep ($bucket[ $_ % $tabsize ] .= "$_\n", keys %FLAG);
  64
  65 my $maxcnt=0;
  66
  67 for ($i=0; $i < $tabsize; $i++)
  68 {
  69     my $cnt=0;
  70
  71     grep ( ++$cnt, split (/\n/, $bucket[$i]));
  72
  73     $maxcnt=$cnt if $cnt > $maxcnt;
  74 }
  75
  76 print "const unsigned unicode_case_hash=$tabsize;\n";
  77 print "/* unicode_case_maxbucket="
  78     . ($maxcnt+2) . "*/\n";
  79
  80 print "const unicode_char unicode_case_tab[][4]={\n";
  81
  82 my $idx=0;
  83
  84 for ($i=0; $i<$tabsize; $i++)
  85 {
  86     $offset[$i]=$idx;
  87
  88     grep {
  89         my $j=$_;
  90         my $u=$UC{$j}+0;
  91         my $l=$LC{$j}+0;
  92         my $t=$TC{$j}+0;
  93
  94         if ($u || $l || $t)
  95         {
  96             $u=$j unless $u;
  97             $l=$j unless $l;
  98             $t=$u unless $t;
  99
 100             printf("{0x%04x,0x%04x,0x%04x,0x%04x},",$j,$u,$l,$t);
 101             print "\n" if ($idx % 4) == 3;
 102             ++$idx;
 103         }
 104     } split(/\n/, $bucket[$i]);
 105 }
 106 print "{0,0,0,0}};
 107
 108 const unsigned unicode_case_offset[$tabsize]={
 109 ";
 110
 111 for ($i=0; $i<$tabsize;$i++)
 112 {
 113     printf("%4d", $offset[$i]);
 114     print "," if $i < $tabsize-1;
 115
 116     print "\n" if ($i % 16) == 15;
 117 }
 118 print "};\n";