IkiWiki/Plugin/htmlbalance.pm

   1 #!/usr/bin/perl
   2 package IkiWiki::Plugin::htmlbalance;
   3
   4 # htmlbalance: Parse and re-serialize HTML to ensure balanced tags
   5 #
   6 # Copyright 2008 Simon McVittie <http://smcv.pseudorandom.co.uk/>
   7 # Licensed under the GNU GPL, version 2, or any later version published by the
   8 # Free Software Foundation
   9
  10 use warnings;
  11 use strict;
  12 use IkiWiki 2.00;
  13 use HTML::TreeBuilder;
  14 use HTML::Entities;
  15
  16 sub import {
  17         hook(type => "getsetup", id => "htmlbalance", call => \&getsetup);
  18         hook(type => "sanitize", id => "htmlbalance", call => \&sanitize);
  19 }
  20
  21 sub getsetup () {
  22         return
  23                 plugin => {
  24                         safe => 1,
  25                         rebuild => undef,
  26                 },
  27 }
  28
  29 sub sanitize (@) {
  30         my %params=@_;
  31         my $ret = '';
  32
  33         my $tree = HTML::TreeBuilder->new();
  34         $tree->ignore_unknown(0);
  35         $tree->ignore_ignorable_whitespace(0);
  36         $tree->no_space_compacting(1);
  37         $tree->p_strict(1);
  38         $tree->store_comments(0);
  39         $tree->store_declarations(0);
  40         $tree->store_pis(0);
  41         $tree->parse_content($params{content});
  42         my @nodes = $tree->disembowel();
  43         foreach my $node (@nodes) {
  44                 if (ref $node) {
  45                         $ret .= $node->as_XML();
  46                         chomp $ret;
  47                         $node->delete();
  48                 }
  49                 else {
  50                         $ret .= encode_entities($node);
  51                 }
  52         }
  53         $tree->delete();
  54         return $ret;
  55 }
  56
  57 1