X-Git-Url: http://git.vanrenterghem.biz/git.ikiwiki.info.git/blobdiff_plain/c5b02885e934e5a1f8c5a8753ed47eaf4cce8b9f..26c2c63c97e6bfa704357503fdfd8d8834838851:/IkiWiki/Plugin/aggregate.pm

diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm
index 898854641..9c28651f0 100644
--- a/IkiWiki/Plugin/aggregate.pm
+++ b/IkiWiki/Plugin/aggregate.pm
@@ -70,11 +70,11 @@ 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} ? $params{dir} : "feed/".IkiWiki::titlepage($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 +83,7 @@ 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;
@@ -93,7 +94,7 @@ sub preprocess (@) { #{{{
 
 	return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
 	       "<i>".$feed->{message}."</i> (".$feed->{numposts}.
-	       " stored posts; ".$feed->{newposts}." new)";
+	       " stored posts; ".$feed->{newposts}." new)<br />";
 } # }}}
 
 sub delete (@) { #{{{
@@ -107,6 +108,8 @@ sub delete (@) { #{{{
 } #}}}
 
 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: $!");
@@ -116,8 +119,9 @@ 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 +144,8 @@ 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 +165,9 @@ 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}};
@@ -180,7 +187,7 @@ sub aggregate () { #{{{
 	eval q{use HTML::Entities};
 	die $@ if $@;
 
-FEED:	foreach my $feed (values %feeds) {
+	foreach my $feed (values %feeds) {
 		next unless time - $feed->{lastupdate} >= $feed->{updateinterval};
 		$feed->{lastupdate}=time;
 		$feed->{newposts}=0;
@@ -188,30 +195,38 @@ FEED:	foreach my $feed (values %feeds) {
 
 		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";
 	}
 
@@ -220,7 +235,7 @@ FEED:	foreach my $feed (values %feeds) {
 
 sub add_page (@) { #{{{
 	my %params=@_;
-
+	
 	my $feed=$params{feed};
 	my $guid={};
 	my $mtime;
@@ -237,7 +252,10 @@ sub add_page (@) { #{{{
 		$feed->{newposts}++;
 
 		# assign it an unused page
-		my $page=$feed->{dir}."/".IkiWiki::titlepage($params{title});
+		my $page=IkiWiki::titlepage($params{title});
+		$page=~s!([/])!"__".ord($1)."__"!eg; # escape slashes in title
+		$page=$feed->{dir}."/".$page;
+		$page=lc($page);
 		($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
 		if (! defined $page || ! length $page) {
 			$page=$feed->{dir}."/item";
@@ -257,7 +275,8 @@ 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});
+	require Encode;
+	my $digest=md5_hex(Encode::encode_utf8($params{content}));
 	return unless ! exists $guid->{md5} || $guid->{md5} ne $digest;
 	$guid->{md5}=$digest;