glib/gen-script-table.pl

   1 #!/usr/bin/perl -w
   2 #
   3 # Script to convert http://www.unicode.org/Public/UNIDATA/Scripts.txt
   4 # into a machine-readable table.
   5 #
   6 ######################################################################
   7
   8 if (@ARGV != 1) {
   9     die "Usage: gen-script-table.pl Scripts.txt > gscripttable.h\n";
  10 }
  11
  12 open IN, $ARGV[0] || die "Cannot open $ARGV[0]: $!\n";
  13
  14 my @ranges;
  15 my $file;
  16 my $easy_range;
  17 my $i;
  18 my $start;
  19 my $end;
  20 my $script;
  21
  22
  23 while (<IN>) {
  24     if (/^\#\s+(Scripts-.*.txt)/) {
  25         $file = $1;
  26     }
  27
  28     s/#.*//;
  29     next if /^\s*$/;
  30     if (!/^([0-9A-F]+)(?:\.\.([0-9A-F]+))?\s*;\s*([A-Za-z_]+)\s*$/) {
  31         die "Cannot parse line: '$_'\n";
  32     }
  33
  34     if (defined $2) {
  35         push @ranges, [ hex $1, hex $2, uc $3 ];
  36     } else {
  37         push @ranges, [ hex $1, hex $1, uc $3 ];
  38     }
  39 }
  40
  41 @ranges = sort { $a->[0] <=> $b->[0] } @ranges;
  42 $date = gmtime;
  43
  44 print <<"EOT";
  45 /* gscripttable.h: Generated by gen-script-table.pl
  46  *
  47  *  Date: $date
  48  *  Source: $file
  49  *
  50  * Do not edit.
  51  */
  52
  53 EOT
  54
  55 $easy_range = 0x2000;
  56
  57 print <<"EOT";
  58 #define G_EASY_SCRIPTS_RANGE $easy_range
  59
  60 static const guchar g_script_easy_table[$easy_range] = {
  61 EOT
  62
  63 $i = 0;
  64 $end = -1;
  65
  66 for (my $c = 0; $c < $easy_range; $c++) {
  67
  68     if ($c % 3 == 0) {
  69       printf "\n ";
  70     }
  71
  72     if ($c > $end) {
  73         $start = $ranges[$i]->[0];
  74         $end = $ranges[$i]->[1];
  75         $script = $ranges[$i]->[2];
  76         $i++;
  77     }
  78
  79     if ($c < $start) {
  80         printf " G_SCRIPT_UNKNOWN,";
  81     } else {
  82         printf " G_SCRIPT_%s,", $script;
  83     }
  84 }
  85
  86 if ($end >= $easy_range) {
  87     $i--;
  88     $ranges[$i]->[0] = $easy_range;
  89 }
  90
  91
  92 print <<"EOT";
  93
  94 };
  95
  96 static const struct {
  97     gunichar    start;
  98     guint16     chars;
  99     guint16     script;
 100 } g_script_table[] = {
 101 EOT
 102
 103 for (; $i <= $#ranges; $i++) {
 104     $start = $ranges[$i]->[0];
 105     $end = $ranges[$i]->[1];
 106     $script = $ranges[$i]->[2];
 107
 108     while ($i <= $#ranges - 1 &&
 109            $ranges[$i + 1]->[0] == $end + 1 &&
 110            $ranges[$i + 1]->[2] eq $script) {
 111         $i++;
 112         $end = $ranges[$i]->[1];
 113     }
 114
 115     printf " { %#06x, %5d, G_SCRIPT_%s },\n", $start, $end - $start + 1, $script;
 116 }
 117
 118 printf "};\n";
 119