Unicode/gen-case

   1 #! /usr/bin/perl
   2 use strict;
   3 use warnings;
   4
   5 my @trans;
   6
   7 print "\t/* -*- c -*- source code generated by ", join(" ", $0, @ARGV), " */\n";
   8 while (<>) {
   9     s/#.*$//;
  10     next if /^\s*$/;
  11     my($code, $status, $mapping) = /^([[:xdigit:]]+);\s*([CFST]);\s*([[:xdigit:]]+(?:\s+[[:xdigit:]]+)*);\s*$/
  12         or warn("$ARGV:$.: weird line\n"), next;
  13     next unless $status eq "C" or $status eq "S";
  14     warn("$ARGV:$.: multi-char simple mapping\n"), next
  15         if $mapping =~ /\s/;
  16     $code = hex($code);
  17     $mapping = hex($mapping);
  18     $trans[$code] = $mapping;
  19 } continue {
  20     close ARGV if eof;
  21 }
  22
  23 sub gobble {
  24     my($begin, $step) = @_;
  25     my $diff = $trans[$begin] - $begin;
  26     my @codes;
  27     my @holes;
  28     my $probe = $begin;
  29     my $hole;
  30     while (1) {
  31         my @beyond;
  32         while (defined($trans[$probe]) && $trans[$probe] == $probe + $diff) {
  33             push @beyond, $probe;
  34             $probe += $step;
  35         }
  36         last unless @beyond >= 2;
  37         push @holes, $hole if defined $hole;
  38         push @codes, @beyond;
  39         $hole = $probe;
  40         $probe += $step;
  41     }
  42     return 0 unless @codes;
  43
  44     # The following formula was tuned for i486-linux-gnu-gcc-4.0 -O1.
  45     if (@codes <= 2 + @holes) {
  46         print "if (", join(" || ", map { sprintf("c == 0x%X", $_) } @codes), ")\n";
  47     } else {
  48         printf "if (c >= 0x%X && c <= 0x%X", $codes[0], $codes[-1];
  49         printf " && c != 0x%X", $_ foreach @holes;
  50         if ($step == 2) { printf " && (c & 1) == %d", $begin & 1 }
  51         elsif ($step != 1) { printf " && c %% %d == %d", $step, $begin % $step }
  52         print ")\n";
  53     }
  54     if ($diff != 0) {
  55         if ($diff < 0) { printf "\t\tc -= "; $diff = -$diff }
  56         else { printf "\t\tc += " }
  57         if ($diff < 10) { printf "%d", $diff }
  58         else { printf "0x%X", $diff }
  59     }
  60     print ";\n";
  61
  62     undef $trans[$_] foreach @codes;
  63     return 1;
  64 }
  65
  66 my $first = 1;
  67 for (my $code = 0; $code <= $#trans; ++$code) {
  68     next unless defined $trans[$code];
  69
  70     print $first ? "\t" : "\telse ";
  71     gobble($code, 1) or gobble($code, 2) or gobble($code, 3) or gobble($code, 4)
  72         or printf "if (c == 0x%X)\n\t\tc = 0x%X;\n", $code, $trans[$code];
  73     $first = 0;
  74 }
  75 close STDOUT or die "$0: -: $!\n";
  76
  77 __END__
  78
  79 =head1 NAME
  80
  81 gen-case - Generate C source code for folding the case of a Unicode character.
  82
  83 =head1 SYNOPSIS
  84
  85 B<gen-case> CaseFolding.txt > ../src/intl/casefold.inc
  86
  87 =head1 DESCRIPTION
  88
  89 B<gen-case> reads F<CaseFolding.txt> of the Unicode Character Database
  90 and generates C source code that implements the I<simple case folding>
  91 as defined in that file.
  92
  93 The generated source code can then be used like this:
  94
  95   unicode_val_T
  96   unicode_simple_case_fold(unicode_val_T c)
  97   {
  98   #include "casefold.inc"
  99           return c;
 100   }
 101
 102 =head1 BUGS
 103
 104 Does not support B<--help> nor B<--version>.
 105
 106 =head1 AUTHOR
 107
 108 Kalle Olavi Niemitalo <kon@iki.fi>
 109
 110 =head1 COPYRIGHT AND LICENSE
 111
 112 Copyright (c) 2006 Kalle Olavi Niemitalo.
 113
 114 This program is free software; you can redistribute it and/or modify
 115 it under the same terms as Perl itself.  In addition:
 116
 117 =over
 118
 119 Permission is hereby granted, free of charge, to any person obtaining a
 120 copy of this software and associated documentation files (the "Software"),
 121 to deal in the Software without restriction, including without limitation
 122 the rights to use, copy, modify, merge, publish, distribute, sublicense,
 123 and/or sell copies of the Software, and to permit persons to whom the
 124 Software is furnished to do so, subject to the following conditions:
 125
 126 The above copyright notice and this permission notice shall be included in
 127 all copies or substantial portions of the Software.
 128
 129 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 130 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 131 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 132 THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 133 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 134 FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 135 DEALINGS IN THE SOFTWARE.
 136
 137 =back