X-Git-Url: http://git.vanrenterghem.biz/git.ikiwiki.info.git/blobdiff_plain/dae0f48e91304afcb6ebe0936360e51b22a56548..de468ba12dfc0891145b3f8b432eb97b70b60c0c:/IkiWiki/Plugin/aggregate.pm

diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm
index ac55dcb1d..b5354a823 100644
--- a/IkiWiki/Plugin/aggregate.pm
+++ b/IkiWiki/Plugin/aggregate.pm
@@ -1,14 +1,15 @@
 #!/usr/bin/perl
-# Blog aggregation plugin.
+# Feed aggregation plugin.
 package IkiWiki::Plugin::aggregate;
 
 use warnings;
 use strict;
-use IkiWiki;
-use HTML::Entities;
+use IkiWiki 2.00;
 use HTML::Parser;
 use HTML::Tagset;
+use HTML::Entities;
 use URI;
+use open qw{:utf8 :std};
 
 my %feeds;
 my %guids;
@@ -16,38 +17,113 @@ my %guids;
 sub import { #{{{
 	hook(type => "getopt", id => "aggregate", call => \&getopt);
 	hook(type => "checkconfig", id => "aggregate", call => \&checkconfig);
-	hook(type => "filter", id => "aggregate", call => \&filter);
+	hook(type => "needsbuild", id => "aggregate", call => \&needsbuild);
 	hook(type => "preprocess", id => "aggregate", call => \&preprocess);
         hook(type => "delete", id => "aggregate", call => \&delete);
 	hook(type => "savestate", id => "aggregate", call => \&savestate);
+	if (exists $config{aggregate_webtrigger} && $config{aggregate_webtrigger}) {
+		hook(type => "cgi", id => "aggregate", call => \&cgi);
+	}
 } # }}}
 
 sub getopt () { #{{{
         eval q{use Getopt::Long};
+	error($@) if $@;
         Getopt::Long::Configure('pass_through');
         GetOptions("aggregate" => \$config{aggregate});
 } #}}}
 
 sub checkconfig () { #{{{
-	IkiWiki::lockwiki();
+	if ($config{aggregate} && ! ($config{post_commit} && 
+	                             IkiWiki::commit_hook_enabled())) {
+		launchaggregation();
+	}
+} #}}}
+
+sub cgi ($) { #{{{
+	my $cgi=shift;
+
+	if (defined $cgi->param('do') &&
+	    $cgi->param("do") eq "aggregate_webtrigger") {
+		$|=1;
+		print "Content-Type: text/plain\n\n";
+		$config{cgi}=0;
+		$config{verbose}=1;
+		$config{syslog}=0;
+		print gettext("Aggregation triggered via web.")."\n\n";
+		if (launchaggregation()) {
+			IkiWiki::lockwiki();
+			IkiWiki::loadindex();
+			require IkiWiki::Render;
+			IkiWiki::refresh();
+			IkiWiki::saveindex();
+		}
+		else {
+			print gettext("Nothing to do right now, all feeds are up-to-date!")."\n";
+		}
+		exit 0;
+	}
+} #}}}
+
+sub launchaggregation () { #{{{
+	# See if any feeds need aggregation.
 	loadstate();
-	if ($config{aggregate}) {
+	my @feeds=needsaggregate();
+	return unless @feeds;
+	if (! lockaggregate()) {
+		debug("an aggregation process is already running");
+		return;
+	}
+	# force a later rebuild of source pages
+	$IkiWiki::forcerebuild{$_->{sourcepage}}=1
+		foreach @feeds;
+
+	# Fork a child process to handle the aggregation.
+	# The parent process will then handle building the
+	# result. This avoids messy code to clear state
+	# accumulated while aggregating.
+	defined(my $pid = fork) or error("Can't fork: $!");
+	if (! $pid) {
 		IkiWiki::loadindex();
-		aggregate();
+		# Aggregation happens without the main wiki lock
+		# being held. This allows editing pages etc while
+		# aggregation is running.
+		aggregate(@feeds);
+
+		IkiWiki::lockwiki;
+		# Merge changes, since aggregation state may have
+		# changed on disk while the aggregation was happening.
+		mergestate();
+		expire();
 		savestate();
+		IkiWiki::unlockwiki;
+		exit 0;
+	}
+	waitpid($pid,0);
+	if ($?) {
+		error "aggregation failed with code $?";
 	}
-	IkiWiki::unlockwiki();
-} #}}}
 
-sub filter (@) { #{{{
-	my %params=@_;
-	my $page=$params{page};
+	clearstate();
+	unlockaggregate();
 
-	# Mark all feeds originating on this page as removable;
-	# preprocess will unmark those that still exist.
-	remove_feeds($page);
+	return 1;
+} #}}}
 
-	return $params{content};
+sub needsbuild (@) { #{{{
+	my $needsbuild=shift;
+	
+	loadstate();
+
+	foreach my $feed (values %feeds) {
+		if (exists $pagesources{$feed->{sourcepage}} && 
+		    grep { $_ eq $pagesources{$feed->{sourcepage}} } @$needsbuild) {
+			# Mark all feeds originating on this page as 
+			# not yet seen; preprocess will unmark those that
+			# still exist.
+			markunseen($feed->{sourcepage});
+		}
+	}
 } # }}}
 
 sub preprocess (@) { #{{{
@@ -55,7 +131,7 @@ sub preprocess (@) { #{{{
 
 	foreach my $required (qw{name url}) {
 		if (! exists $params{$required}) {
-			return "[[aggregate plugin missing $required parameter]]";
+			return "[[aggregate ".sprintf(gettext("missing %s parameter"), $required)."]]";
 		}
 	}
 
@@ -78,11 +154,19 @@ sub preprocess (@) { #{{{
 	$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;
-	delete $feed->{remove};
+        if (exists $params{template}) {
+                $params{template}=~s/[^-_a-zA-Z0-9]+//g;
+        }
+        else {
+                $params{template} = "aggregatepost"
+        }
+	$feed->{template}=$params{template} . ".tmpl";
+	delete $feed->{unseen};
 	$feed->{lastupdate}=0 unless defined $feed->{lastupdate};
 	$feed->{numposts}=0 unless defined $feed->{numposts};
 	$feed->{newposts}=0 unless defined $feed->{newposts};
-	$feed->{message}="new feed" unless defined $feed->{message};
+	$feed->{message}=gettext("new feed") unless defined $feed->{message};
+	$feed->{error}=0 unless defined $feed->{error};
 	$feed->{tags}=[];
 	while (@_) {
 		my $key=shift;
@@ -93,8 +177,11 @@ sub preprocess (@) { #{{{
 	}
 
 	return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
-	       "<i>".$feed->{message}."</i> (".$feed->{numposts}." posts".
-	       ($feed->{newposts} ? "; ".$feed->{newposts}." new" : "").
+	       ($feed->{error} ? "<em>" : "").$feed->{message}.
+	       ($feed->{error} ? "</em>" : "").
+	       " (".$feed->{numposts}." ".gettext("posts").
+	       ($feed->{newposts} ? "; ".$feed->{newposts}.
+	                            " ".gettext("new") : "").
 	       ")";
 } # }}}
 
@@ -104,14 +191,28 @@ sub delete (@) { #{{{
 	# Remove feed data for removed pages.
 	foreach my $file (@files) {
 		my $page=pagename($file);
-		remove_feeds($page);
+		markunseen($page);
 	}
 } #}}}
 
+sub markunseen ($) { #{{{
+	my $page=shift;
+
+	foreach my $id (keys %feeds) {
+		if ($feeds{$id}->{sourcepage} eq $page) {
+			$feeds{$id}->{unseen}=1;
+		}
+	}
+} #}}}
+
+my $state_loaded=0;
+
 sub loadstate () { #{{{
+	return if $state_loaded;
+	$state_loaded=1;
 	if (-e "$config{wikistatedir}/aggregate") {
-		open (IN, "$config{wikistatedir}/aggregate" ||
-			die "$config{wikistatedir}/aggregate: $!");
+		open(IN, "$config{wikistatedir}/aggregate") ||
+			die "$config{wikistatedir}/aggregate: $!";
 		while (<IN>) {
 			$_=IkiWiki::possibly_foolish_untaint($_);
 			chomp;
@@ -143,25 +244,12 @@ sub loadstate () { #{{{
 } #}}}
 
 sub savestate () { #{{{
-	eval q{use HTML::Entities};
-	die $@ if $@;
-	open (OUT, ">$config{wikistatedir}/aggregate" ||
-		die "$config{wikistatedir}/aggregate: $!");
+	return unless $state_loaded;
+	garbage_collect();
+	my $newfile="$config{wikistatedir}/aggregate.new";
+	my $cleanup = sub { unlink($newfile) };
+	open (OUT, ">$newfile") || error("open $newfile: $!", $cleanup);
 	foreach my $data (values %feeds, values %guids) {
-		if ($data->{remove}) {
-			if ($data->{name}) {
-				foreach my $guid (values %guids) {
-					if ($guid->{feed} eq $data->{name}) {
-						$guid->{remove}=1;
-					}
-				}
-			}
-			else {
-				unlink pagefile($data->{page});
-			}
-			next;
-		}
-
 		my @line;
 		foreach my $field (keys %$data) {
 			if ($field eq "name" || $field eq "feed" ||
@@ -175,43 +263,176 @@ sub savestate () { #{{{
 				push @line, "$field=".$data->{$field};
 			}
 		}
-		print OUT join(" ", @line)."\n";
+		print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
 	}
-	close OUT;
+	close OUT || error("save $newfile: $!", $cleanup);
+	rename($newfile, "$config{wikistatedir}/aggregate") ||
+		error("rename $newfile: $!", $cleanup);
 } #}}}
 
-sub aggregate () { #{{{
-	eval q{use XML::Feed};
-	die $@ if $@;
-	eval q{use HTML::Entities};
-	die $@ if $@;
+sub garbage_collect () { #{{{
+	foreach my $name (keys %feeds) {
+		# remove any feeds that were not seen while building the pages
+		# that used to contain them
+		if ($feeds{$name}->{unseen}) {
+			delete $feeds{$name};
+		}
+	}
+
+	foreach my $guid (values %guids) {
+		# any guid whose feed is gone should be removed
+		if (! exists $feeds{$guid->{feed}}) {
+			unlink pagefile($guid->{page})
+				if exists $guid->{page};
+			delete $guids{$guid->{guid}};
+		}
+		# handle expired guids
+		elsif ($guid->{expired} && exists $guid->{page}) {
+			unlink pagefile($guid->{page});
+			delete $guid->{page};
+			delete $guid->{md5};
+		}
+	}
+} #}}}
+
+sub mergestate () { #{{{
+	# Load the current state in from disk, and merge into it
+	# values from the state in memory that might have changed
+	# during aggregation.
+	my %myfeeds=%feeds;
+	my %myguids=%guids;
+	clearstate();
+	loadstate();
+
+	# All that can change in feed state during aggregation is a few
+	# fields.
+	foreach my $name (keys %myfeeds) {
+		if (exists $feeds{$name}) {
+			foreach my $field (qw{message lastupdate numposts
+			                      newposts error}) {
+				$feeds{$name}->{$field}=$myfeeds{$name}->{$field};
+			}
+		}
+	}
+
+	# New guids can be created during aggregation.
+	# It's also possible that guids were removed from the on-disk state
+	# while the aggregation was in process. That would only happen if
+	# their feed was also removed, so any removed guids added back here
+	# will be garbage collected later.
+	foreach my $guid (keys %myguids) {
+		if (! exists $guids{$guid}) {
+			$guids{$guid}=$myguids{$guid};
+		}
+	}
+} #}}}
+
+sub clearstate () { #{{{
+	%feeds=();
+	%guids=();
+	$state_loaded=0;
+} #}}}
 
+sub expire () { #{{{
 	foreach my $feed (values %feeds) {
-		next unless $config{rebuild} || 
-			time - $feed->{lastupdate} >= $feed->{updateinterval};
+		next unless $feed->{expireage} || $feed->{expirecount};
+		my $count=0;
+		my %seen;
+		foreach my $item (sort { $IkiWiki::pagectime{$b->{page}} <=> $IkiWiki::pagectime{$a->{page}} }
+		                  grep { exists $_->{page} && $_->{feed} eq $feed->{name} && $IkiWiki::pagectime{$_->{page}} }
+		                  values %guids) {
+			if ($feed->{expireage}) {
+				my $days_old = (time - $IkiWiki::pagectime{$item->{page}}) / 60 / 60 / 24;
+				if ($days_old > $feed->{expireage}) {
+					debug(sprintf(gettext("expiring %s (%s days old)"),
+						$item->{page}, int($days_old)));
+					$item->{expired}=1;
+				}
+			}
+			elsif ($feed->{expirecount} &&
+			       $count >= $feed->{expirecount}) {
+				debug(sprintf(gettext("expiring %s"), $item->{page}));
+				$item->{expired}=1;
+			}
+			else {
+				if (! $seen{$item->{page}}) {
+					$seen{$item->{page}}=1;
+					$count++;
+				}
+			}
+		}
+	}
+} #}}}
+
+sub needsaggregate () { #{{{
+	return values %feeds if $config{rebuild};
+	return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds;
+} #}}}
+
+sub aggregate (@) { #{{{
+	eval q{use XML::Feed};
+	error($@) if $@;
+	eval q{use URI::Fetch};
+	error($@) if $@;
+
+	foreach my $feed (@_) {
 		$feed->{lastupdate}=time;
 		$feed->{newposts}=0;
-		$IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
+		$feed->{message}=sprintf(gettext("processed ok at %s"),
+			displaytime($feed->{lastupdate}));
+		$feed->{error}=0;
 
-		debug("checking feed ".$feed->{name}." ...");
+		debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
 
 		if (! length $feed->{feedurl}) {
 			my @urls=XML::Feed->find_feeds($feed->{url});
 			if (! @urls) {
-				$feed->{message}="could not find feed at ".$feed->{feedurl};
+				$feed->{message}=sprintf(gettext("could not find feed at %s"), $feed->{url});
+				$feed->{error}=1;
 				debug($feed->{message});
 				next;
 			}
 			$feed->{feedurl}=pop @urls;
 		}
-		my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))};
+		my $res=URI::Fetch->fetch($feed->{feedurl});
+		if (! $res) {
+			$feed->{message}=URI::Fetch->errstr;
+			$feed->{error}=1;
+			debug($feed->{message});
+			next;
+		}
+		if ($res->status == URI::Fetch::URI_GONE()) {
+			$feed->{message}=gettext("feed not found");
+			$feed->{error}=1;
+			debug($feed->{message});
+			next;
+		}
+		my $content=$res->content;
+		my $f=eval{XML::Feed->parse(\$content)};
+		if ($@) {
+			# One common cause of XML::Feed crashing is a feed
+			# that contains invalid UTF-8 sequences. Convert
+			# feed to ascii to try to work around.
+			$feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)"));
+			$content=Encode::decode_utf8($content, 0);
+			$f=eval{XML::Feed->parse(\$content)};
+		}
 		if ($@) {
-			$feed->{message}="feed crashed XML::Feed! $@";
+			# Another possibility is badly escaped entities.
+			$feed->{message}.=" ".sprintf(gettext("(feed entities escaped)"));
+			$content=~s/\&(?!amp)(\w+);/&amp;$1;/g;
+			$content=Encode::decode_utf8($content, 0);
+			$f=eval{XML::Feed->parse(\$content)};
+		}
+		if ($@) {
+			$feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
+			$feed->{error}=1;
 			debug($feed->{message});
 			next;
 		}
 		if (! $f) {
 			$feed->{message}=XML::Feed->errstr;
+			$feed->{error}=1;
 			debug($feed->{message});
 			next;
 		}
@@ -219,19 +440,15 @@ sub aggregate () { #{{{
 		foreach my $entry ($f->entries) {
 			add_page(
 				feed => $feed,
+				copyright => $f->copyright,
 				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,
+				content => defined $entry->content->body ? $entry->content->body : "",
+				guid => defined $entry->id ? $entry->id : time."_".$feed->{name},
 				ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
 			);
 		}
-
-		$feed->{message}="processed ok at ".
-			displaytime($feed->{lastupdate});
 	}
-
-	# TODO: expiry
 } #}}}
 
 sub add_page (@) { #{{{
@@ -243,6 +460,7 @@ sub add_page (@) { #{{{
 	if (exists $guids{$params{guid}}) {
 		# updating an existing post
 		$guid=$guids{$params{guid}};
+		return if $guid->{expired};
 	}
 	else {
 		# new post
@@ -267,8 +485,21 @@ sub add_page (@) { #{{{
 		       -e pagefile($page.$c)) {
 			$c++
 		}
+
+		# Make sure that the file name isn't too long. 
+		# NB: This doesn't check for path length limits.
+		my $max=POSIX::pathconf($config{srcdir}, &POSIX::_PC_NAME_MAX);
+		if (defined $max && length(htmlfn($page)) >= $max) {
+			$c="";
+			$page=$feed->{dir}."/item";
+			while (exists $IkiWiki::pagecase{lc $page.$c} ||
+			       -e pagefile($page.$c)) {
+				$c++
+			}
+		}
+
 		$guid->{page}=$page;
-		debug("creating new page $page");
+		debug(sprintf(gettext("creating new page %s"), $page));
 	}
 	$guid->{feed}=$feed->{name};
 	
@@ -276,29 +507,33 @@ 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'};
+	error($@) if $@;
 	require Encode;
 	my $digest=md5_hex(Encode::encode_utf8($params{content}));
 	return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
 	$guid->{md5}=$digest;
 
 	# Create the page.
-	my $template=template("aggregatepost.tmpl", blind_cache => 1);
+	my $template=template($feed->{template}, blind_cache => 1);
 	$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(url => $feed->{url});
+	$template->param(copyright => $params{copyright})
+		if defined $params{copyright} && length $params{copyright};
 	$template->param(permalink => urlabs($params{link}, $feed->{feedurl}))
 		if defined $params{link};
 	if (ref $feed->{tags}) {
 		$template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
 	}
-	writefile(htmlpage($guid->{page}), $config{srcdir},
+	writefile(htmlfn($guid->{page}), $config{srcdir},
 		$template->output);
 
 	# Set the mtime, this lets the build process get the right creation
 	# time on record for the new page.
-	utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
+	utime $mtime, $mtime, pagefile($guid->{page})
+		if defined $mtime && $mtime <= time;
 } #}}}
 
 sub htmlescape ($) { #{{{
@@ -353,22 +588,36 @@ sub htmlabs ($$) { #{{{
 	return $ret;
 } #}}}
 
-sub remove_feeds () { #{{{
+sub pagefile ($) { #{{{
 	my $page=shift;
 
-	my %removed;
-	foreach my $id (keys %feeds) {
-		if ($feeds{$id}->{sourcepage} eq $page) {
-			$feeds{$id}->{remove}=1;
-			$removed{$id}=1;
-		}
-	}
+	return "$config{srcdir}/".htmlfn($page);
 } #}}}
 
-sub pagefile ($) { #{{{
-	my $page=shift;
+sub htmlfn ($) { #{{{
+	return shift().".".$config{htmlext};
+} #}}}
+
+my $aggregatelock;
+
+sub lockaggregate () { #{{{
+	# Take an exclusive lock to prevent multiple concurrent aggregators.
+	# Returns true if the lock was aquired.
+	if (! -d $config{wikistatedir}) {
+		mkdir($config{wikistatedir});
+	}
+	open($aggregatelock, '>', "$config{wikistatedir}/aggregatelock") ||
+		error ("cannot open to $config{wikistatedir}/aggregatelock: $!");
+	if (! flock($aggregatelock, 2 | 4)) { # LOCK_EX | LOCK_NB
+		close($aggregatelock) || error("failed closing aggregatelock: $!");
+		return 0;
+	}
+	return 1;
+} #}}}
 
-	return "$config{srcdir}/".htmlpage($page);
+sub unlockaggregate () { #{{{
+	return close($aggregatelock) if $aggregatelock;
+	return;
 } #}}}
 
 1