libs/unicode/mkgraphemebreak.pl

   1 #! /usr/bin/perl
   2 #
   3 # Compile GraphemeBreak.txt into C array declarations.
   4 #
   5 # The array's structure is [firstchar, lastchar, class], giving the
   6 # linebreaking "class" for unicode character range firstchar-lastchar.
   7 #
   8 # The ranges are sorted in numerical order.
   9 #
  10 # An array gets generated for each block of 4096 unicode characters.
  11 #
  12 # Finally, two arrays get declared: a pointer to an array for each 4096
  13 # unicode character block, and the number of elements in the array.
  14 #
  15 # The pointer is NULL for each block of 4096 unicode characters that is not
  16 # defined in LineBreak.txt
  17 #
  18 # By definition, a unicode character that is not listed in the array is
  19 # class XX.
  20
  21 use strict;
  22 use warnings;
  23 use mkcommon;
  24
  25 open(F, "<GraphemeBreakProperty.txt") || die;
  26
  27 # Read LineBreak.txt
  28 #
  29 # Invoke doclass() passing the first unicode char, the last unicode char,
  30 # and the linebreaking class for the given unicode char range.
  31
  32 my @breaklist;
  33
  34 while (defined($_=<F>))
  35 {
  36     chomp;
  37
  38     next unless /^([0-9A-F]+)(\.\.([0-9A-F]+))?\s*\;\s*([^\s]+)\s*/;
  39
  40     my $f=$1;
  41     my $l=$3;
  42     my $t=$4;
  43
  44     $l=$f unless $l;
  45
  46     eval "\$f=0x$f";
  47     eval "\$l=0x$l";
  48
  49     push @breaklist, [$f, $l, $t];
  50 }
  51
  52
  53 my $obj=mkcommon->new;
  54
  55 foreach (sort { $$a[0] <=> $$b[0] } @breaklist)
  56 {
  57     $obj->range($$_[0], $$_[1], "UNICODE_GRAPHEMEBREAK_$$_[2]");
  58 }
  59
  60 $obj->output;