* Encode link and guid urls in rss feeds to avoid illegal utf8 slipping in.

[git.ikiwiki.info.git] / IkiWiki / Plugin / aggregate.pm

diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm
index 16db42a8fbd4f917d3cb4ac75313988295dbc558..7c4ab3d106b87bf5cc4ce834bad15e7aaba7c7e1 100644 (file)
--- a/IkiWiki/Plugin/aggregate.pm
+++ b/IkiWiki/Plugin/aggregate.pm
@@ -6,6 +6,9 @@ use warnings;
 use strict;
 use IkiWiki;
 use HTML::Entities;
+use HTML::Parser;
+use HTML::Tagset;
+use URI;
 
 my %feeds;
 my %guids;
@@ -283,15 +286,13 @@ sub add_page (@) { #{{{
 
        # Create the page.
        my $template=IkiWiki::template("aggregatepost.tmpl", blind_cache => 1);
-       my $content=$params{content};
-       $params{content}=~s/(?<!\\)\[\[/\\\[\[/g; # escape accidental wikilinks
-                                                 # and preprocessor stuff
        $template->param(title => $params{title})
                if defined $params{title} && length($params{title});
-       $template->param(content => $params{content});
-       $template->param(url => $feed->{url});
+       $template->param(content => htmlescape(htmlabs($params{content}, $feed->{feedurl})));
        $template->param(name => $feed->{name});
-       $template->param(link => $params{link}) if defined $params{link};
+       $template->param(url => $feed->{url});
+       $template->param(permalink => urlabs($params{link}, $feed->{feedurl}))
+               if defined $params{link};
        if (ref $feed->{tags}) {
                $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
        }
@@ -303,6 +304,58 @@ sub add_page (@) { #{{{
        utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
 } #}}}
 
+sub htmlescape ($) { #{{{
+       # escape accidental wikilinks and preprocessor stuff
+       my $html=shift;
+       $html=~s/(?<!\\)\[\[/\\\[\[/g;
+       return $html;
+} #}}}
+
+sub urlabs ($$) { #{{{
+       my $url=shift;
+       my $urlbase=shift;
+
+       URI->new_abs($url, $urlbase)->as_string;
+} #}}}
+
+sub htmlabs ($$) { #{{{
+       # Convert links in html from relative to absolute.
+       # Note that this is a heuristic, which is not specified by the rss
+       # spec and may not be right for all feeds. Also, see Debian
+       # bug #XXXX TODO: get bug.
+       my $html=shift;
+       my $urlbase=shift;
+
+       my $ret="";
+       my $p = HTML::Parser->new(api_version => 3);
+       $p->handler(default => sub { $ret.=join("", @_) }, "text");
+       $p->handler(start => sub {
+               my ($tagname, $pos, $text) = @_;
+               if (ref $HTML::Tagset::linkElements{$tagname}) {
+                       while (4 <= @$pos) {
+                               # use attribute sets from right to left
+                               # to avoid invalidating the offsets
+                               # when replacing the values
+                               my($k_offset, $k_len, $v_offset, $v_len) =
+                                       splice(@$pos, -4);
+                               my $attrname = lc(substr($text, $k_offset, $k_len));
+                               next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
+                               next unless $v_offset; # 0 v_offset means no value
+                               my $v = substr($text, $v_offset, $v_len);
+                               $v =~ s/^([\'\"])(.*)\1$/$2/;
+                               my $new_v=urlabs($v, $urlbase);
+                               $new_v =~ s/\"/&quot;/g; # since we quote with ""
+                               substr($text, $v_offset, $v_len) = qq("$new_v");
+                       }
+               }
+               $ret.=$text;
+       }, "tagname, tokenpos, text");
+       $p->parse($html);
+       $p->eof;
+
+       return $ret;
+} #}}}
+
 sub remove_feeds () { #{{{
        my $page=shift;