]> sipb.mit.edu Git - ikiwiki.git/blobdiff - IkiWiki/Plugin/aggregate.pm
add aggregate locking functions
[ikiwiki.git] / IkiWiki / Plugin / aggregate.pm
index 5024975ac8603223191b9a36525c6a650d6bec77..cfc4ec955328e2dfa812b04b4cacd683ba336884 100644 (file)
@@ -17,7 +17,7 @@ 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);
@@ -33,8 +33,10 @@ sub getopt () { #{{{
 sub checkconfig () { #{{{
        if ($config{aggregate} && ! ($config{post_commit} && 
                                     IkiWiki::commit_hook_enabled())) {
-               # don't wait for the lock
-               IkiWiki::lockwiki(0) || exit 1;
+               if (! IkiWiki::lockwiki(0)) {
+                       debug("wiki is locked by another process, not aggregating");
+                       exit 1;
+               }
        
                loadstate();
                IkiWiki::loadindex();
@@ -47,16 +49,19 @@ sub checkconfig () { #{{{
        }
 } #}}}
 
-sub filter (@) { #{{{
-       my %params=@_;
-       my $page=$params{page};
-
+sub needsbuild (@) { #{{{
+       my $needsbuild=shift;
+       
        loadstate(); # if not already loaded
-       # Mark all feeds originating on this page as removable;
-       # preprocess will unmark those that still exist.
-       remove_feeds($page);
 
-       return $params{content};
+       foreach my $feed (values %feeds) {
+               if (exists $pagesources{$feed->{sourcepage}} && 
+                   grep { $_ eq $pagesources{$feed->{sourcepage}} } @$needsbuild) {
+                       # Mark all feeds originating on this page as removable;
+                       # preprocess will unmark those that still exist.
+                       remove_feeds($feed->{sourcepage});
+               }
+       }
 } # }}}
 
 sub preprocess (@) { #{{{
@@ -125,9 +130,10 @@ sub delete (@) { #{{{
 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;
@@ -155,19 +161,16 @@ sub loadstate () { #{{{
                }
 
                close IN;
-               
-               $state_loaded=1;
        }
 } #}}}
 
 sub savestate () { #{{{
+       return unless $state_loaded;
        eval q{use HTML::Entities};
        error($@) if $@;
        my $newfile="$config{wikistatedir}/aggregate.new";
-       # TODO: This cleanup function could use improvement. Any newly
-       # aggregated files are left behind unrecorded, and should be deleted.
        my $cleanup = sub { unlink($newfile) };
-       open (OUT, ">$newfile") || error("open $newfile: $!", $cleanup);
+       open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup);
        foreach my $data (values %feeds, values %guids) {
                if ($data->{remove}) {
                        if ($data->{name}) {
@@ -178,7 +181,8 @@ sub savestate () { #{{{
                                }
                        }
                        else {
-                               unlink pagefile($data->{page});
+                               unlink pagefile($data->{page})
+                                       if exists $data->{page};
                        }
                        next;
                }
@@ -218,6 +222,7 @@ sub expire () { #{{{
        foreach my $feed (values %feeds) {
                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) {
@@ -225,7 +230,7 @@ sub expire () { #{{{
                                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}, $days_old));
+                                               $item->{page}, int($days_old)));
                                        $item->{expired}=1;
                                }
                        }
@@ -235,7 +240,10 @@ sub expire () { #{{{
                                $item->{expired}=1;
                        }
                        else {
-                               $count++;
+                               if (! $seen{$item->{page}}) {
+                                       $seen{$item->{page}}=1;
+                                       $count++;
+                               }
                        }
                }
        }
@@ -294,6 +302,13 @@ sub aggregate () { #{{{
                        $content=Encode::decode_utf8($content);
                        $f=eval{XML::Feed->parse(\$content)};
                }
+               if ($@) {
+                       # 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);
+                       $f=eval{XML::Feed->parse(\$content)};
+               }
                if ($@) {
                        $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
                        $feed->{error}=1;
@@ -310,6 +325,7 @@ 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 => defined $entry->content->body ? $entry->content->body : "",
@@ -389,6 +405,8 @@ sub add_page (@) { #{{{
        $template->param(content => htmlescape(htmlabs($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}) {
@@ -399,7 +417,8 @@ sub add_page (@) { #{{{
 
        # 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 ($) { #{{{
@@ -473,7 +492,29 @@ sub pagefile ($) { #{{{
 } #}}}
 
 sub htmlfn ($) { #{{{
-       return shift().".html";
+       return shift().".".$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: $!");
+               return 0;
+       }
+       return 1;
+} #}}}
+
+sub unlockaggregate () { #{{{
+       return close($aggregatelock) if $aggregatelock;
+       return;
 } #}}}
 
 1