misc/translator/text-extract.pl

   1 #!/usr/bin/perl
   2 #use strict;
   3 #use warnings; FIXME - Bug 2505
   4 use HTML::Tree;
   5 use Getopt::Std;
   6 getopt("f:");
   7         my $tree = HTML::TreeBuilder->new; # empty tree
   8
   9         $tree->parse_file($opt_f);
  10         sub give_id {
  11                 my $x = $_[0];
  12                 foreach my $c ($x->content_list) {
  13                         next if (ref($c) && $c->tag() eq "~comment");
  14                         next if (ref($c) && $c->tag() eq "script");
  15                         next if (ref($c) && $c->tag() eq "style");
  16                         if (!ref($c)) {
  17                                 print "$c\n";
  18                         }
  19                         if (ref($c) && $c->attr('alt')) {
  20                                 print $c->attr('alt')."\n";
  21                         }
  22                         if (ref($c) && $c->attr('title')) {
  23                                 print $c->attr('title')."\n";
  24                         }
  25                         if (ref($c) && $c->tag() eq "input" && $c->attr('value')) {
  26                                 print $c->attr('value')."\n";
  27                         }
  28                         if (ref($c) && $c->tag() eq 'meta') {
  29                                 print $c->attr('content')."\n ";
  30                         }
  31                         give_id($c) if ref $c; # ignore text nodes
  32                 }
  33         };
  34         give_id($tree);
  35         $tree = $tree->delete;