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 3.00;
  13 use HTML::Entities;
  14
  15 sub import {
  16         hook(type => "getsetup", id => "htmlbalance", call => \&getsetup);
  17         hook(type => "sanitize", id => "htmlbalance", call => \&sanitize);
  18 }
  19
  20 sub getsetup () {
  21         return
  22                 plugin => {
  23                         safe => 1,
  24                         rebuild => undef,
  25                 },
  26 }
  27
  28 sub sanitize (@) {
  29         my %params=@_;
  30         my $ret = '';
  31
  32         eval q{use HTML::TreeBuilder};
  33         error $@ if $@;
  34         my $tree = HTML::TreeBuilder->new();
  35         $tree->ignore_unknown(0);
  36         $tree->ignore_ignorable_whitespace(0);
  37         $tree->no_space_compacting(1);
  38         $tree->p_strict(1);
  39         $tree->store_comments(0);
  40         $tree->store_declarations(0);
  41         $tree->store_pis(0);
  42         $tree->parse_content($params{content});
  43         my @nodes = $tree->disembowel();
  44         foreach my $node (@nodes) {
  45                 if (ref $node) {
  46                         $ret .= $node->as_XML();
  47                         chomp $ret;
  48                         $node->delete();
  49                 }
  50                 else {
  51                         $ret .= encode_entities($node);
  52                 }
  53         }
  54         $tree->delete();
  55         return $ret;
  56 }
  57
  58 1