IkiWiki/Plugin/htmlscrubber.pm

   1 #!/usr/bin/perl
   2 package IkiWiki::Plugin::htmlscrubber;
   3
   4 use warnings;
   5 use strict;
   6 use IkiWiki 2.00;
   7
   8 sub import { #{{{
   9         hook(type => "sanitize", id => "htmlscrubber", call => \&sanitize);
  10 } # }}}
  11
  12 sub sanitize (@) { #{{{
  13         my %params=@_;
  14         return scrubber()->scrub($params{content});
  15 } # }}}
  16
  17 my $_scrubber;
  18 sub scrubber { #{{{
  19         return $_scrubber if defined $_scrubber;
  20
  21         # Only known uri schemes are allowed to avoid all the ways of
  22         # embedding javascrpt.
  23         # List at http://en.wikipedia.org/wiki/URI_scheme
  24         my $uri_schemes=join("|",
  25                 # IANA registered schemes
  26                 "http", "https", "ftp", "mailto", "file", "telnet", "gopher",
  27                 "aaa", "aaas", "acap",  "cap", "cid", "crid",
  28                 "dav", "dict", "dns", "fax", "go", "h323", "im", "imap",
  29                 "ldap", "mid", "news", "nfs", "nntp", "pop", "pres",
  30                 "sip", "sips", "snmp", "tel", "urn", "wais", "xmpp",
  31                 "z39.50r", "z39.50s",
  32                 # Selected unofficial schemes
  33                 "aim", "callto", "cvs", "ed2k", "feed", "fish", "gg",
  34                 "irc", "ircs", "lastfm", "ldaps", "magnet", "mms",
  35                 "msnim", "notes", "rsync", "secondlife", "skype", "ssh",
  36                 "sftp", "sms", "steam", "webcal", "ymsgr",
  37         );
  38         # data is a special case. Allow data:image/*, but
  39         # disallow data:text/javascript and everything else.
  40         my $link=qr/^(?:(?:$uri_schemes):|data:image\/|[^:]+$)/i;
  41
  42         eval q{use HTML::Scrubber};
  43         error($@) if $@;
  44         # Lists based on http://feedparser.org/docs/html-sanitization.html
  45         # With html 5 video and audio tags added.
  46         $_scrubber = HTML::Scrubber->new(
  47                 allow => [qw{
  48                         a abbr acronym address area b big blockquote br br/
  49                         button caption center cite code col colgroup dd del
  50                         dfn dir div dl dt em fieldset font form h1 h2 h3 h4
  51                         h5 h6 hr hr/ i img input ins kbd label legend li map
  52                         menu ol optgroup option p p/ pre q s samp select small
  53                         span strike strong sub sup table tbody td textarea
  54                         tfoot th thead tr tt u ul var
  55                         video audio
  56                 }],
  57                 default => [undef, { (
  58                         map { $_ => 1 } qw{
  59                                 abbr accept accept-charset accesskey
  60                                 align alt axis border cellpadding cellspacing
  61                                 char charoff charset checked cite class
  62                                 clear cols colspan color compact coords
  63                                 datetime dir disabled enctype for frame
  64                                 headers height hreflang hspace id ismap
  65                                 label lang longdesc maxlength media method
  66                                 multiple name nohref noshade nowrap prompt
  67                                 readonly rel rev rows rowspan rules scope
  68                                 selected shape size span start summary
  69                                 tabindex target title type usemap valign
  70                                 value vspace width
  71                                 autoplay loopstart loopend end
  72                                 playcount controls
  73                         } ),
  74                         "/" => 1, # emit proper <hr /> XHTML
  75                         href => $link,
  76                         src => $link,
  77                         action => $link,
  78                         poster => $link,
  79                 }],
  80         );
  81         return $_scrubber;
  82 } # }}}
  83
  84 1