libs/unicode/mkultcase.pl

   1 # Copyright 2000-2004 Double Precision, Inc.
   2 # See COPYING for distribution information.
   3 #
   4 #
   5 # Generate unicode upper/lower/titlecase translations.
   6
   7 print '/*
   8 ** Copyright 2000-2004 Double Precision, Inc.
   9 ** See COPYING for distribution information.
  10 **
  11 */
  12
  13 #include "unicode.h"
  14
  15 ';
  16
  17 my $set=shift;
  18
  19 open (U, "UnicodeData.txt") || die "$!\n";
  20
  21 while (<U>)
  22 {
  23         chomp;
  24
  25 my @fields= split /;/;
  26
  27 my ($code, $uc, $lc, $tc);
  28
  29         $code="0x$fields[0]";
  30         eval "\$code=$code;";
  31
  32         $uc=$fields[12];
  33         if ($uc ne "")
  34         {
  35                 eval "\$uc=0x$uc;";
  36                 $UC{$code}=$uc;
  37                 $FLAG{$code}=1;
  38         }
  39
  40         $lc=$fields[13];
  41         if ($lc ne "")
  42         {
  43                 eval "\$lc=0x$lc;";
  44                 $LC{$code}=$lc;
  45                 $FLAG{$code}=1;
  46         }
  47
  48         $tc=$fields[14];
  49         if ($tc ne "")
  50         {
  51                 eval "\$tc=0x$tc;";
  52                 $TC{$code}=$tc;
  53                 $FLAG{$code}=1;
  54         }
  55 }
  56
  57 close(U);
  58
  59 my $tabsize=2048;
  60
  61 grep ($bucket[ $_ % $tabsize ] .= "$_\n", keys %FLAG);
  62
  63 my $maxcnt=0;
  64
  65 for ($i=0; $i < $tabsize; $i++)
  66 {
  67     my $cnt=0;
  68
  69     grep ( ++$cnt, split (/\n/, $bucket[$i]));
  70
  71     $maxcnt=$cnt if $cnt > $maxcnt;
  72 }
  73
  74 print "const unsigned unicode_case_hash=$tabsize;\n";
  75 print "/* unicode_case_maxbucket="
  76     . ($maxcnt+2) . "*/\n";
  77
  78 print "const unicode_char unicode_case_tab[][4]={\n";
  79
  80 my $idx=0;
  81
  82 for ($i=0; $i<$tabsize; $i++)
  83 {
  84     $offset[$i]=$idx;
  85
  86     grep {
  87         my $j=$_;
  88         my $u=$UC{$j}+0;
  89         my $l=$LC{$j}+0;
  90         my $t=$TC{$j}+0;
  91
  92         if ($u || $l || $t)
  93         {
  94             $u=$j unless $u;
  95             $l=$j unless $l;
  96             $t=$u unless $t;
  97
  98             printf("{0x%04x,0x%04x,0x%04x,0x%04x},",$j,$u,$l,$t);
  99             print "\n" if ($idx % 4) == 3;
 100             ++$idx;
 101         }
 102     } split(/\n/, $bucket[$i]);
 103 }
 104 print "{0,0,0,0}};
 105
 106 const unsigned unicode_case_offset[$tabsize]={
 107 ";
 108
 109 for ($i=0; $i<$tabsize;$i++)
 110 {
 111     printf("%4d", $offset[$i]);
 112     print "," if $i < $tabsize-1;
 113
 114     print "\n" if ($i % 16) == 15;
 115 }
 116 print "};\n";