misc/translator/text-extract.pl

   1 #!/usr/bin/perl
   2 #use strict;
   3 #use warnings; FIXME - Bug 2505
   4
   5 use FindBin;
   6 use lib $FindBin::Bin;
   7
   8 use HTML::Tree;
   9 use Getopt::Std;
  10 getopt("f:");
  11         my $tree = HTML::TreeBuilder->new; # empty tree
  12
  13         $tree->parse_file($opt_f);
  14         sub give_id {
  15                 my $x = $_[0];
  16                 foreach my $c ($x->content_list) {
  17                         next if (ref($c) && $c->tag() eq "~comment");
  18                         next if (ref($c) && $c->tag() eq "script");
  19                         next if (ref($c) && $c->tag() eq "style");
  20                         if (!ref($c)) {
  21                                 print "$c\n";
  22                         }
  23                         if (ref($c) && $c->attr('alt')) {
  24                                 print $c->attr('alt')."\n";
  25                         }
  26                         if (ref($c) && $c->attr('title')) {
  27                                 print $c->attr('title')."\n";
  28                         }
  29                         if (ref($c) && $c->tag() eq "input" && $c->attr('value')) {
  30                                 print $c->attr('value')."\n";
  31                         }
  32                         if (ref($c) && $c->tag() eq 'meta') {
  33                                 print $c->attr('content')."\n ";
  34                         }
  35                         give_id($c) if ref $c; # ignore text nodes
  36                 }
  37         };
  38         give_id($tree);
  39         $tree = $tree->delete;