use warnings;
use strict;
use IkiWiki;
+use HTML::Entities;
+use HTML::Parser;
+use HTML::Tagset;
+use URI;
my %feeds;
my %guids;
$feed->{name}=$name;
$feed->{sourcepage}=$params{page};
$feed->{url}=$params{url};
- my $dir=exists $params{dir} ? $params{dir} : IkiWiki::titlepage($params{name});
+ my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".IkiWiki::titlepage($params{name});
$dir=~s/^\/+//;
($dir)=$dir=~/$IkiWiki::config{wiki_file_regexp}/;
$feed->{dir}=$dir;
} #}}}
sub loadstate () { #{{{
- eval q{use HTML::Entities};
- die $@ if $@;
if (-e "$IkiWiki::config{wikistatedir}/aggregate") {
open (IN, "$IkiWiki::config{wikistatedir}/aggregate" ||
die "$IkiWiki::config{wikistatedir}/aggregate: $!");
die $@ if $@;
foreach my $feed (values %feeds) {
- next unless time - $feed->{lastupdate} >= $feed->{updateinterval};
+ next unless $IkiWiki::config{rebuild} ||
+ time - $feed->{lastupdate} >= $feed->{updateinterval};
$feed->{lastupdate}=time;
$feed->{newposts}=0;
$IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
$feed->{newposts}++;
# assign it an unused page
- my $page=$feed->{dir}."/".IkiWiki::titlepage($params{title});
+ my $page=IkiWiki::titlepage($params{title});
+ # escape slashes and periods in title so it doesn't specify
+ # directory name or trigger ".." disallowing code.
+ $page=~s!([/.])!"__".ord($1)."__"!eg;
+ $page=$feed->{dir}."/".$page;
$page=lc($page);
($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
if (! defined $page || ! length $page) {
$page=$feed->{dir}."/item";
}
- $page=~s/\.\.//g; # avoid ".." directory tricks
my $c="";
while (exists $IkiWiki::pagesources{$page.$c} ||
-e pagefile($page.$c)) {
eval q{use Digest::MD5 'md5_hex'};
require Encode;
my $digest=md5_hex(Encode::encode_utf8($params{content}));
- return unless ! exists $guid->{md5} || $guid->{md5} ne $digest;
+ return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $IkiWiki::config{rebuild};
$guid->{md5}=$digest;
# 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(content => $params{content});
- $template->param(url => $feed->{url});
+ $template->param(title => $params{title})
+ if defined $params{title} && length($params{title});
+ $template->param(content => htmlescape(htmlabs($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}}]);
}
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);
+ $new_v =~ s/\"/"/g; # since we quote with ""
+ substr($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;