2 # Feed aggregation plugin.
3 package IkiWiki::Plugin::aggregate;
11 use open qw{:utf8 :std};
17 hook(type => "getopt", id => "aggregate", call => \&getopt);
18 hook(type => "getsetup", id => "aggregate", call => \&getsetup);
19 hook(type => "checkconfig", id => "aggregate", call => \&checkconfig,
21 hook(type => "needsbuild", id => "aggregate", call => \&needsbuild);
22 hook(type => "preprocess", id => "aggregate", call => \&preprocess);
23 hook(type => "delete", id => "aggregate", call => \&delete);
24 hook(type => "savestate", id => "aggregate", call => \&savestate);
25 hook(type => "htmlize", id => "_aggregated", call => \&htmlize);
26 if (exists $config{aggregate_webtrigger} && $config{aggregate_webtrigger}) {
27 hook(type => "cgi", id => "aggregate", call => \&cgi);
32 eval q{use Getopt::Long};
34 Getopt::Long::Configure('pass_through');
36 "aggregate" => \$config{aggregate},
37 "aggregateinternal!" => \$config{aggregateinternal},
47 aggregateinternal => {
50 description => "enable aggregation to internal pages?",
51 safe => 0, # enabling needs manual transition
54 aggregate_webtrigger => {
57 description => "allow aggregation to be triggered via the web?",
63 example => { file => "$ENV{HOME}/.ikiwiki/cookies" },
64 safe => 0, # hooks into perl module internals
65 description => "cookie control",
70 if (! defined $config{aggregateinternal}) {
71 $config{aggregateinternal}=1;
73 if (! defined $config{cookiejar}) {
74 $config{cookiejar}={ file => "$ENV{HOME}/.ikiwiki/cookies" };
77 # This is done here rather than in a refresh hook because it
78 # needs to run before the wiki is locked.
79 if ($config{aggregate} && ! ($config{post_commit} &&
80 IkiWiki::commit_hook_enabled())) {
88 if (defined $cgi->param('do') &&
89 $cgi->param("do") eq "aggregate_webtrigger") {
91 print "Content-Type: text/plain\n\n";
95 print gettext("Aggregation triggered via web.")."\n\n";
96 if (launchaggregation()) {
99 require IkiWiki::Render;
101 IkiWiki::saveindex();
104 print gettext("Nothing to do right now, all feeds are up-to-date!")."\n";
110 sub launchaggregation () {
111 # See if any feeds need aggregation.
113 my @feeds=needsaggregate();
114 return unless @feeds;
115 if (! lockaggregate()) {
116 error("an aggregation process is already running");
118 # force a later rebuild of source pages
119 $IkiWiki::forcerebuild{$_->{sourcepage}}=1
122 # Fork a child process to handle the aggregation.
123 # The parent process will then handle building the
124 # result. This avoids messy code to clear state
125 # accumulated while aggregating.
126 defined(my $pid = fork) or error("Can't fork: $!");
128 IkiWiki::loadindex();
129 # Aggregation happens without the main wiki lock
130 # being held. This allows editing pages etc while
131 # aggregation is running.
135 # Merge changes, since aggregation state may have
136 # changed on disk while the aggregation was happening.
145 error "aggregation failed with code $?";
154 # Pages with extension _aggregated have plain html markup, pass through.
157 return $params{content};
160 # Used by ikiwiki-transition aggregateinternal.
161 sub migrate_to_internal {
162 if (! lockaggregate()) {
163 error("an aggregation process is currently running");
170 foreach my $data (values %guids) {
171 next unless $data->{page};
172 next if $data->{expired};
174 $config{aggregateinternal} = 0;
175 my $oldname = "$config{srcdir}/".htmlfn($data->{page});
177 $oldname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page});
180 my $oldoutput = $config{destdir}."/".IkiWiki::htmlpage($data->{page});
182 $config{aggregateinternal} = 1;
183 my $newname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page});
185 debug "moving $oldname -> $newname";
188 error("$newname already exists");
191 debug("already renamed to $newname?");
194 elsif (-e $oldname) {
195 rename($oldname, $newname) || error("$!");
198 debug("$oldname not found");
201 require IkiWiki::Render;
202 debug("removing output file $oldoutput");
203 IkiWiki::prune($oldoutput, $config{destdir});
214 my $needsbuild=shift;
218 foreach my $feed (values %feeds) {
219 if (exists $pagesources{$feed->{sourcepage}} &&
220 grep { $_ eq $pagesources{$feed->{sourcepage}} } @$needsbuild) {
221 # Mark all feeds originating on this page as
222 # not yet seen; preprocess will unmark those that
224 markunseen($feed->{sourcepage});
234 foreach my $required (qw{name url}) {
235 if (! exists $params{$required}) {
236 error sprintf(gettext("missing %s parameter"), $required)
241 my $name=$params{name};
242 if (exists $feeds{$name}) {
249 $feed->{sourcepage}=$params{page};
250 $feed->{url}=$params{url};
251 my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".titlepage($params{name});
253 ($dir)=$dir=~/$config{wiki_file_regexp}/;
255 $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
256 $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
257 $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
258 $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
259 if (exists $params{template}) {
260 $params{template}=~s/[^-_a-zA-Z0-9]+//g;
263 $params{template} = "aggregatepost"
265 $feed->{template}=$params{template} . ".tmpl";
266 delete $feed->{unseen};
267 $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
268 $feed->{lasttry}=$feed->{lastupdate} unless defined $feed->{lasttry};
269 $feed->{numposts}=0 unless defined $feed->{numposts};
270 $feed->{newposts}=0 unless defined $feed->{newposts};
271 $feed->{message}=gettext("new feed") unless defined $feed->{message};
272 $feed->{error}=0 unless defined $feed->{error};
278 push @{$feed->{tags}}, $value;
282 return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
283 ($feed->{error} ? "<em>" : "").$feed->{message}.
284 ($feed->{error} ? "</em>" : "").
285 " (".$feed->{numposts}." ".gettext("posts").
286 ($feed->{newposts} ? "; ".$feed->{newposts}.
287 " ".gettext("new") : "").
294 # Remove feed data for removed pages.
295 foreach my $file (@files) {
296 my $page=pagename($file);
304 foreach my $id (keys %feeds) {
305 if ($feeds{$id}->{sourcepage} eq $page) {
306 $feeds{$id}->{unseen}=1;
314 return if $state_loaded;
316 if (-e "$config{wikistatedir}/aggregate") {
317 open(IN, "<", "$config{wikistatedir}/aggregate") ||
318 die "$config{wikistatedir}/aggregate: $!";
320 $_=IkiWiki::possibly_foolish_untaint($_);
323 foreach my $i (split(/ /, $_)) {
324 my ($field, $val)=split(/=/, $i, 2);
325 if ($field eq "name" || $field eq "feed" ||
326 $field eq "guid" || $field eq "message") {
327 $data->{$field}=decode_entities($val, " \t\n");
329 elsif ($field eq "tag") {
330 push @{$data->{tags}}, $val;
333 $data->{$field}=$val;
337 if (exists $data->{name}) {
338 $feeds{$data->{name}}=$data;
340 elsif (exists $data->{guid}) {
341 $guids{$data->{guid}}=$data;
350 return unless $state_loaded;
352 my $newfile="$config{wikistatedir}/aggregate.new";
353 my $cleanup = sub { unlink($newfile) };
354 open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup);
355 foreach my $data (values %feeds, values %guids) {
357 foreach my $field (keys %$data) {
358 if ($field eq "name" || $field eq "feed" ||
359 $field eq "guid" || $field eq "message") {
360 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
362 elsif ($field eq "tags") {
363 push @line, "tag=$_" foreach @{$data->{tags}};
366 push @line, "$field=".$data->{$field}
367 if defined $data->{$field};
370 print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
372 close OUT || error("save $newfile: $!", $cleanup);
373 rename($newfile, "$config{wikistatedir}/aggregate") ||
374 error("rename $newfile: $!", $cleanup);
377 foreach my $feed (keys %feeds) {
378 my $t=$feeds{$feed}->{lastupdate}+$feeds{$feed}->{updateinterval};
379 if (! defined $timestamp || $timestamp > $t) {
383 $newfile=~s/\.new$/time/;
384 open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup);
385 if (defined $timestamp) {
386 print OUT $timestamp."\n";
388 close OUT || error("save $newfile: $!", $cleanup);
391 sub garbage_collect () {
392 foreach my $name (keys %feeds) {
393 # remove any feeds that were not seen while building the pages
394 # that used to contain them
395 if ($feeds{$name}->{unseen}) {
396 delete $feeds{$name};
400 foreach my $guid (values %guids) {
401 # any guid whose feed is gone should be removed
402 if (! exists $feeds{$guid->{feed}}) {
403 if (exists $guid->{page}) {
404 unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page})
405 || unlink "$config{srcdir}/".htmlfn($guid->{page});
407 delete $guids{$guid->{guid}};
409 # handle expired guids
410 elsif ($guid->{expired} && exists $guid->{page}) {
411 unlink "$config{srcdir}/".htmlfn($guid->{page});
412 unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page});
413 delete $guid->{page};
420 # Load the current state in from disk, and merge into it
421 # values from the state in memory that might have changed
422 # during aggregation.
428 # All that can change in feed state during aggregation is a few
430 foreach my $name (keys %myfeeds) {
431 if (exists $feeds{$name}) {
432 foreach my $field (qw{message lastupdate lasttry
433 numposts newposts error}) {
434 $feeds{$name}->{$field}=$myfeeds{$name}->{$field};
439 # New guids can be created during aggregation.
440 # Guids have a few fields that may be updated during aggregation.
441 # It's also possible that guids were removed from the on-disk state
442 # while the aggregation was in process. That would only happen if
443 # their feed was also removed, so any removed guids added back here
444 # will be garbage collected later.
445 foreach my $guid (keys %myguids) {
446 if (! exists $guids{$guid}) {
447 $guids{$guid}=$myguids{$guid};
450 foreach my $field (qw{md5}) {
451 $guids{$guid}->{$field}=$myguids{$guid}->{$field};
464 foreach my $feed (values %feeds) {
465 next unless $feed->{expireage} || $feed->{expirecount};
468 foreach my $item (sort { ($IkiWiki::pagectime{$b->{page}} || 0) <=> ($IkiWiki::pagectime{$a->{page}} || 0) }
469 grep { exists $_->{page} && $_->{feed} eq $feed->{name} }
471 if ($feed->{expireage}) {
472 my $days_old = (time - ($IkiWiki::pagectime{$item->{page}} || 0)) / 60 / 60 / 24;
473 if ($days_old > $feed->{expireage}) {
474 debug(sprintf(gettext("expiring %s (%s days old)"),
475 $item->{page}, int($days_old)));
479 elsif ($feed->{expirecount} &&
480 $count >= $feed->{expirecount}) {
481 debug(sprintf(gettext("expiring %s"), $item->{page}));
485 if (! $seen{$item->{page}}) {
486 $seen{$item->{page}}=1;
494 sub needsaggregate () {
495 return values %feeds if $config{rebuild};
496 return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds;
500 eval q{use Net::INET6Glue::INET_is_INET6}; # may not be available
501 eval q{use XML::Feed};
503 eval q{use URI::Fetch};
506 foreach my $feed (@_) {
507 $feed->{lasttry}=time;
509 $feed->{message}=sprintf(gettext("last checked %s"),
510 displaytime($feed->{lasttry}));
513 debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
515 if (! length $feed->{feedurl}) {
516 my @urls=XML::Feed->find_feeds($feed->{url});
518 $feed->{message}=sprintf(gettext("could not find feed at %s"), $feed->{url});
520 debug($feed->{message});
523 $feed->{feedurl}=pop @urls;
525 my $res=URI::Fetch->fetch($feed->{feedurl},
526 UserAgent => LWP::UserAgent->new(
527 cookie_jar => $config{cookiejar},
531 $feed->{message}=URI::Fetch->errstr;
533 debug($feed->{message});
537 # lastupdate is only set if we were able to contact the server
538 $feed->{lastupdate}=$feed->{lasttry};
540 if ($res->status == URI::Fetch::URI_GONE()) {
541 $feed->{message}=gettext("feed not found");
543 debug($feed->{message});
546 my $content=$res->content;
547 my $f=eval{XML::Feed->parse(\$content)};
549 # One common cause of XML::Feed crashing is a feed
550 # that contains invalid UTF-8 sequences. Convert
551 # feed to ascii to try to work around.
552 $feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)"));
554 $content=Encode::decode_utf8($content, 0);
555 XML::Feed->parse(\$content)
559 # Another possibility is badly escaped entities.
560 $feed->{message}.=" ".sprintf(gettext("(feed entities escaped)"));
561 $content=~s/\&(?!amp)(\w+);/&$1;/g;
563 $content=Encode::decode_utf8($content, 0);
564 XML::Feed->parse(\$content)
568 $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
570 debug($feed->{message});
574 $feed->{message}=XML::Feed->errstr;
576 debug($feed->{message});
580 foreach my $entry ($f->entries) {
581 # XML::Feed doesn't work around XML::Atom's bizarre
582 # API, so we will. Real unicode strings? Yes please.
583 # See [[bugs/Aggregated_Atom_feeds_are_double-encoded]]
584 local $XML::Atom::ForceUnicode = 1;
586 my $c=$entry->content;
587 # atom feeds may have no content, only a summary
588 if (! defined $c && ref $entry->summary) {
594 copyright => $f->copyright,
595 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
596 link => $entry->link,
597 content => (defined $c && defined $c->body) ? $c->body : "",
598 guid => defined $entry->id ? $entry->id : time."_".$feed->{name},
599 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
600 base => (defined $c && $c->can("base")) ? $c->base : undef,
609 my $feed=$params{feed};
612 if (exists $guids{$params{guid}}) {
613 # updating an existing post
614 $guid=$guids{$params{guid}};
615 return if $guid->{expired};
616 write_page($feed, $guid, $mtime, \%params);
620 $guid->{guid}=$params{guid};
621 $guids{$params{guid}}=$guid;
622 $mtime=$params{ctime};
626 # assign it an unused page
627 my $page=titlepage($params{title});
628 # escape slashes and periods in title so it doesn't specify
629 # directory name or trigger ".." disallowing code.
630 $page=~s!([/.])!"__".ord($1)."__"!eg;
631 $page=$feed->{dir}."/".$page;
632 ($page)=$page=~/$config{wiki_file_regexp}/;
633 if (! defined $page || ! length $page) {
634 $page=$feed->{dir}."/item";
637 while (exists $IkiWiki::pagecase{lc $page.$c} ||
638 -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) ||
639 -e "$config{srcdir}/".htmlfn($page.$c)) {
644 eval { write_page($feed, $guid, $mtime, \%params) };
646 # assume failure was due to a too long filename
649 $page=$feed->{dir}."/item";
650 while (exists $IkiWiki::pagecase{lc $page.$c} ||
651 -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) ||
652 -e "$config{srcdir}/".htmlfn($page.$c)) {
657 write_page($feed, $guid, $mtime, \%params);
660 debug(sprintf(gettext("creating new page %s"), $page));
664 sub write_page ($$$$$) {
668 my %params=%{shift()};
670 $guid->{feed}=$feed->{name};
672 # To write or not to write? Need to avoid writing unchanged pages
673 # to avoid unneccessary rebuilding. The mtime from rss cannot be
674 # trusted; let's use a digest.
675 eval q{use Digest::MD5 'md5_hex'};
678 my $digest=md5_hex(Encode::encode_utf8($params{content}));
679 return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
680 $guid->{md5}=$digest;
685 $template=template($feed->{template}, blind_cache => 1);
688 print STDERR gettext("failed to process template:")." $@";
691 $template->param(title => $params{title})
692 if defined $params{title} && length($params{title});
693 $template->param(content => wikiescape(htmlabs($params{content},
694 defined $params{base} ? $params{base} : $feed->{feedurl})));
695 $template->param(name => $feed->{name});
696 $template->param(url => $feed->{url});
697 $template->param(copyright => $params{copyright})
698 if defined $params{copyright} && length $params{copyright};
699 $template->param(permalink => IkiWiki::urlabs($params{link}, $feed->{feedurl}))
700 if defined $params{link};
701 if (ref $feed->{tags}) {
702 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
704 writefile(htmlfn($guid->{page}),
705 $IkiWiki::Plugin::transient::transientdir, $template->output);
707 if (defined $mtime && $mtime <= time) {
708 # Set the mtime, this lets the build process get the right
709 # creation time on record for the new page.
710 utime $mtime, $mtime,
711 $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page});
712 # Store it in pagectime for expiry code to use also.
713 $IkiWiki::pagectime{$guid->{page}}=$mtime
714 unless exists $IkiWiki::pagectime{$guid->{page}};
717 # Dummy value for expiry code.
718 $IkiWiki::pagectime{$guid->{page}}=time
719 unless exists $IkiWiki::pagectime{$guid->{page}};
724 # escape accidental wikilinks and preprocessor stuff
725 return encode_entities(shift, '\[\]');
729 # Convert links in html from relative to absolute.
730 # Note that this is a heuristic, which is not specified by the rss
731 # spec and may not be right for all feeds. Also, see Debian
737 my $p = HTML::Parser->new(api_version => 3);
738 $p->handler(default => sub { $ret.=join("", @_) }, "text");
739 $p->handler(start => sub {
740 my ($tagname, $pos, $text) = @_;
741 if (ref $HTML::Tagset::linkElements{$tagname}) {
743 # use attribute sets from right to left
744 # to avoid invalidating the offsets
745 # when replacing the values
746 my($k_offset, $k_len, $v_offset, $v_len) =
748 my $attrname = lc(substr($text, $k_offset, $k_len));
749 next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
750 next unless $v_offset; # 0 v_offset means no value
751 my $v = substr($text, $v_offset, $v_len);
752 $v =~ s/^([\'\"])(.*)\1$/$2/;
753 my $new_v=IkiWiki::urlabs($v, $urlbase);
754 $new_v =~ s/\"/"/g; # since we quote with ""
755 substr($text, $v_offset, $v_len) = qq("$new_v");
759 }, "tagname, tokenpos, text");
767 return shift().".".($config{aggregateinternal} ? "_aggregated" : $config{htmlext});
772 sub lockaggregate () {
773 # Take an exclusive lock to prevent multiple concurrent aggregators.
774 # Returns true if the lock was aquired.
775 if (! -d $config{wikistatedir}) {
776 mkdir($config{wikistatedir});
778 open($aggregatelock, '>', "$config{wikistatedir}/aggregatelock") ||
779 error ("cannot open to $config{wikistatedir}/aggregatelock: $!");
780 if (! flock($aggregatelock, 2 | 4)) { # LOCK_EX | LOCK_NB
781 close($aggregatelock) || error("failed closing aggregatelock: $!");
787 sub unlockaggregate () {
788 return close($aggregatelock) if $aggregatelock;