contrib/less/mkutable

   1 #! /usr/bin/perl
   2 use strict;
   3
   4 my $USAGE = <<__EOF__;
   5    usage: mkutable [-n] [-f#] type... [--] [<] UnicodeData.txt
   6           -n = take non-matching types
   7           -f = zero-based type field (default 2)
   8 __EOF__
   9
  10 use vars qw( $opt_f $opt_n );
  11 use Getopt::Std;
  12 my $type_field = 2;
  13
  14 exit (main() ? 1 : 0);
  15
  16 sub main {
  17     my $date = `date`;
  18     chomp $date;
  19     my $args = join ' ', @ARGV;
  20     my $header = "/* Generated by \"$0 $args\" on $date */\n";
  21
  22     die $USAGE if not getopts('f:n');
  23     $type_field = $opt_f if $opt_f;
  24     my %types;
  25     my $arg;
  26     while ($arg = shift @ARGV) {
  27         last if $arg eq '--';
  28         $types{$arg} = 1;
  29     }
  30     my %out = ( 'types' => \%types );
  31     my $last_code = 0;
  32
  33     print $header;
  34     while (<>) {
  35         chomp;
  36         s/#.*//;
  37         my @fields = split /;/;
  38         next if not @fields;
  39         my $code = hex $fields[0];
  40         my $type = $fields[$type_field];
  41         $type =~ s/\s//g;
  42         while (++$last_code < $code) {
  43             output(\%out, $last_code, '?');
  44         }
  45         output(\%out, $code, $type);
  46     }
  47     output(\%out, $last_code+1, '?');
  48 }
  49
  50 sub output {
  51     my ($out, $code, $type) = @_;
  52     my $match = ${${$out}{types}}{$type};
  53     my $type_change = (not $$out{start_type} or $type ne $$out{start_type});
  54     $match = not $match if $opt_n;
  55     if ($match and (not $$out{in_run} or $type_change)) {
  56         end_run($out, $code-1);
  57         start_run($out, $code, $type);
  58     } elsif (not $match and $$out{in_run}) {
  59         end_run($out, $code-1);
  60     }
  61 }
  62
  63 sub start_run {
  64     my ($out, $code, $type) = @_;
  65     $$out{start_code} = $code;
  66     $$out{start_type} = $type;
  67     $$out{in_run} = 1;
  68 }
  69
  70 sub end_run {
  71     my ($out, $code) = @_;
  72     return if not $$out{in_run};
  73     printf "\t{ 0x%04x, 0x%04x }, /* %s */\n", $$out{start_code}, $code, $$out{start_type};
  74     $$out{in_run} = 0;
  75 }