X-Git-Url: http://git.vanrenterghem.biz/git.ikiwiki.info.git/blobdiff_plain/c5b02885e934e5a1f8c5a8753ed47eaf4cce8b9f..22b414ddc0d0e35ad94c60c96a8ac2c8c5e267d5:/IkiWiki/Plugin/aggregate.pm

diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm
index 898854641..d0d7f4217 100644
--- a/IkiWiki/Plugin/aggregate.pm
+++ b/IkiWiki/Plugin/aggregate.pm
@@ -5,6 +5,10 @@ package IkiWiki::Plugin::aggregate;
 use warnings;
 use strict;
 use IkiWiki;
+use HTML::Entities;
+use HTML::Parser;
+use HTML::Tagset;
+use URI;
 
 my %feeds;
 my %guids;
@@ -31,12 +35,14 @@ sub getopt () { #{{{
 } #}}}
 
 sub checkconfig () { #{{{
+	IkiWiki::lockwiki();
 	loadstate();
 	if ($IkiWiki::config{aggregate}) {
 		IkiWiki::loadindex();
 		aggregate();
 		savestate();
 	}
+	IkiWiki::unlockwiki();
 } #}}}
 
 sub filter (@) { #{{{
@@ -70,11 +76,11 @@ sub preprocess (@) { #{{{
 	$feed->{name}=$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;
-	$feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : $params{url};
+	$feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
 	$feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
 	$feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
 	$feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
@@ -83,6 +89,7 @@ sub preprocess (@) { #{{{
 	$feed->{numposts}=0 unless defined $feed->{numposts};
 	$feed->{newposts}=0 unless defined $feed->{newposts};
 	$feed->{message}="new feed" unless defined $feed->{message};
+	$feed->{tags}=[];
 	while (@_) {
 		my $key=shift;
 		my $value=shift;
@@ -116,8 +123,9 @@ sub loadstate () { #{{{
 			my $data={};
 			foreach my $i (split(/ /, $_)) {
 				my ($field, $val)=split(/=/, $i, 2);
-				if ($field eq "name" || $field eq "message") {
-					$data->{$field}=IkiWiki::pagetitle($val);
+				if ($field eq "name" || $field eq "feed" ||
+				    $field eq "guid" || $field eq "message") {
+					$data->{$field}=decode_entities($val, " \t\n");
 				}
 				elsif ($field eq "tag") {
 					push @{$data->{tags}}, $val;
@@ -140,6 +148,8 @@ sub loadstate () { #{{{
 } #}}}
 
 sub savestate () { #{{{
+	eval q{use HTML::Entities};
+	die $@ if $@;
 	open (OUT, ">$IkiWiki::config{wikistatedir}/aggregate" ||
 		die "$IkiWiki::config{wikistatedir}/aggregate: $!");
 	foreach my $data (values %feeds, values %guids) {
@@ -159,8 +169,9 @@ sub savestate () { #{{{
 
 		my @line;
 		foreach my $field (keys %$data) {
-			if ($field eq "name" || $field eq "message") {
-				push @line, "$field=".IkiWiki::titlepage($data->{$field});
+			if ($field eq "name" || $field eq "feed" ||
+			    $field eq "guid" || $field eq "message") {
+				push @line, "$field=".encode_entities($data->{$field}, " \t\n");
 			}
 			elsif ($field eq "tags") {
 				push @line, "tag=$_" foreach @{$data->{tags}};
@@ -180,39 +191,49 @@ sub aggregate () { #{{{
 	eval q{use HTML::Entities};
 	die $@ if $@;
 
-FEED:	foreach my $feed (values %feeds) {
-		next unless time - $feed->{lastupdate} >= $feed->{updateinterval};
+	foreach my $feed (values %feeds) {
+		next unless $IkiWiki::config{rebuild} || 
+			time - $feed->{lastupdate} >= $feed->{updateinterval};
 		$feed->{lastupdate}=time;
 		$feed->{newposts}=0;
 		$IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
 
 		IkiWiki::debug("checking feed ".$feed->{name}." ...");
 
-		my @urls=XML::Feed->find_feeds($feed->{feedurl});
-		if (! @urls) {
-			$feed->{message}="could not find feed at ".$feed->{feedurl};
-			IkiWiki::debug($feed->{message});
-		}
-		foreach my $url (@urls) {
-			my $f=XML::Feed->parse(URI->new($url));
-			if (! $f) {
-				$feed->{message}=XML::Feed->errstr;
+		if (! length $feed->{feedurl}) {
+			my @urls=XML::Feed->find_feeds($feed->{url});
+			if (! @urls) {
+				$feed->{message}="could not find feed at ".$feed->{feedurl};
 				IkiWiki::debug($feed->{message});
-				next FEED;
+				next;
 			}
+			$feed->{feedurl}=pop @urls;
+		}
+		my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))};
+		if ($@) {
+			$feed->{message}="feed crashed XML::Feed! $@";
+			IkiWiki::debug($feed->{message});
+			next;
+		}
+		if (! $f) {
+			$feed->{message}=XML::Feed->errstr;
+			IkiWiki::debug($feed->{message});
+			next;
+		}
 
-			foreach my $entry ($f->entries) {
-				add_page(
-					feed => $feed,
-					title => decode_entities($entry->title),
-					link => $entry->link,
-					content => $entry->content->body,
-					guid => defined $entry->id ? $entry->id : time."_".$feed->name,
-					ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
-				);
-			}
+		foreach my $entry ($f->entries) {
+			add_page(
+				feed => $feed,
+				title => defined $entry->title ? decode_entities($entry->title) : "untitled",
+				link => $entry->link,
+				content => $entry->content->body,
+				guid => defined $entry->id ? $entry->id : time."_".$feed->name,
+				ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
+			);
 		}
-		$feed->{message}="processed ok";
+
+		$feed->{message}="processed ok at ".
+			IkiWiki::displaytime($feed->{lastupdate});
 	}
 
 	# TODO: expiry
@@ -220,7 +241,7 @@ FEED:	foreach my $feed (values %feeds) {
 
 sub add_page (@) { #{{{
 	my %params=@_;
-
+	
 	my $feed=$params{feed};
 	my $guid={};
 	my $mtime;
@@ -237,14 +258,17 @@ sub add_page (@) { #{{{
 		$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)=$page=~/$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} ||
+		while (exists $IkiWiki::pagecase{lc $page.$c} ||
 		       -e pagefile($page.$c)) {
 			$c++
 		}
@@ -257,23 +281,24 @@ sub add_page (@) { #{{{
 	# to avoid unneccessary rebuilding. The mtime from rss cannot be
 	# trusted; let's use a digest.
 	eval q{use Digest::MD5 'md5_hex'};
-	my $digest=md5_hex($params{content});
-	return unless ! exists $guid->{md5} || $guid->{md5} ne $digest;
+	require Encode;
+	my $digest=md5_hex(Encode::encode_utf8($params{content}));
+	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}}]);
 	}
-	IkiWiki::writefile($guid->{page}.".html", $IkiWiki::config{srcdir},
+	IkiWiki::writefile(IkiWiki::htmlpage($guid->{page}), $IkiWiki::config{srcdir},
 		$template->output);
 
 	# Set the mtime, this lets the build process get the right creation
@@ -281,6 +306,58 @@ 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 #381359.
+	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;
 
@@ -296,7 +373,7 @@ sub remove_feeds () { #{{{
 sub pagefile ($) { #{{{
 	my $page=shift;
 
-	return "$IkiWiki::config{srcdir}/$page.html";
+	return "$IkiWiki::config{srcdir}/".IkiWiki::htmlpage($page);
 } #}}}
 
 1