Put newly aggregated pages in the transient underlay
[ikiwiki.git] / IkiWiki / Plugin / aggregate.pm
1 #!/usr/bin/perl
2 # Feed aggregation plugin.
3 package IkiWiki::Plugin::aggregate;
4
5 use warnings;
6 use strict;
7 use IkiWiki 3.00;
8 use HTML::Parser;
9 use HTML::Tagset;
10 use HTML::Entities;
11 use open qw{:utf8 :std};
12
13 my %feeds;
14 my %guids;
15
16 sub import {
17         hook(type => "getopt", id => "aggregate", call => \&getopt);
18         hook(type => "getsetup", id => "aggregate", call => \&getsetup);
19         hook(type => "checkconfig", id => "aggregate", call => \&checkconfig);
20         hook(type => "needsbuild", id => "aggregate", call => \&needsbuild);
21         hook(type => "preprocess", id => "aggregate", call => \&preprocess);
22         hook(type => "delete", id => "aggregate", call => \&delete);
23         hook(type => "savestate", id => "aggregate", call => \&savestate);
24         hook(type => "htmlize", id => "_aggregated", call => \&htmlize);
25         if (exists $config{aggregate_webtrigger} && $config{aggregate_webtrigger}) {
26                 hook(type => "cgi", id => "aggregate", call => \&cgi);
27         }
28 }
29
30 sub getopt () {
31         eval q{use Getopt::Long};
32         error($@) if $@;
33         Getopt::Long::Configure('pass_through');
34         GetOptions(
35                 "aggregate" => \$config{aggregate},
36                 "aggregateinternal!" => \$config{aggregateinternal},
37         );
38 }
39
40 sub getsetup () {
41         return
42                 plugin => {
43                         safe => 1,
44                         rebuild => undef,
45                 },
46                 aggregateinternal => {
47                         type => "boolean",
48                         example => 1,
49                         description => "enable aggregation to internal pages?",
50                         safe => 0, # enabling needs manual transition
51                         rebuild => 0,
52                 },
53                 aggregate_webtrigger => {
54                         type => "boolean",
55                         example => 0,
56                         description => "allow aggregation to be triggered via the web?",
57                         safe => 1,
58                         rebuild => 0,
59                 },
60 }
61
62 sub checkconfig () {
63         if (! defined $config{aggregateinternal}) {
64                 $config{aggregateinternal}=1;
65         }
66
67         if ($config{aggregate} && ! ($config{post_commit} && 
68                                      IkiWiki::commit_hook_enabled())) {
69                 launchaggregation();
70         }
71 }
72
73 sub cgi ($) {
74         my $cgi=shift;
75
76         if (defined $cgi->param('do') &&
77             $cgi->param("do") eq "aggregate_webtrigger") {
78                 $|=1;
79                 print "Content-Type: text/plain\n\n";
80                 $config{cgi}=0;
81                 $config{verbose}=1;
82                 $config{syslog}=0;
83                 print gettext("Aggregation triggered via web.")."\n\n";
84                 if (launchaggregation()) {
85                         IkiWiki::lockwiki();
86                         IkiWiki::loadindex();
87                         require IkiWiki::Render;
88                         IkiWiki::refresh();
89                         IkiWiki::saveindex();
90                 }
91                 else {
92                         print gettext("Nothing to do right now, all feeds are up-to-date!")."\n";
93                 }
94                 exit 0;
95         }
96 }
97
98 sub launchaggregation () {
99         # See if any feeds need aggregation.
100         loadstate();
101         my @feeds=needsaggregate();
102         return unless @feeds;
103         if (! lockaggregate()) {
104                 debug("an aggregation process is already running");
105                 return;
106         }
107         # force a later rebuild of source pages
108         $IkiWiki::forcerebuild{$_->{sourcepage}}=1
109                 foreach @feeds;
110
111         # Fork a child process to handle the aggregation.
112         # The parent process will then handle building the
113         # result. This avoids messy code to clear state
114         # accumulated while aggregating.
115         defined(my $pid = fork) or error("Can't fork: $!");
116         if (! $pid) {
117                 IkiWiki::loadindex();
118                 # Aggregation happens without the main wiki lock
119                 # being held. This allows editing pages etc while
120                 # aggregation is running.
121                 aggregate(@feeds);
122
123                 IkiWiki::lockwiki;
124                 # Merge changes, since aggregation state may have
125                 # changed on disk while the aggregation was happening.
126                 mergestate();
127                 expire();
128                 savestate();
129                 IkiWiki::unlockwiki;
130                 exit 0;
131         }
132         waitpid($pid,0);
133         if ($?) {
134                 error "aggregation failed with code $?";
135         }
136
137         clearstate();
138         unlockaggregate();
139
140         return 1;
141 }
142
143 #  Pages with extension _aggregated have plain html markup, pass through.
144 sub htmlize (@) {
145         my %params=@_;
146         return $params{content};
147 }
148
149 # Used by ikiwiki-transition aggregateinternal.
150 sub migrate_to_internal {
151         if (! lockaggregate()) {
152                 error("an aggregation process is currently running");
153         }
154
155         IkiWiki::lockwiki();
156         loadstate();
157         $config{verbose}=1;
158
159         foreach my $data (values %guids) {
160                 next unless $data->{page};
161                 next if $data->{expired};
162                 
163                 $config{aggregateinternal} = 0;
164                 my $oldname = "$config{srcdir}/".htmlfn($data->{page});
165                 if (! -e $oldname) {
166                         $oldname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page});
167                 }
168
169                 my $oldoutput = $config{destdir}."/".IkiWiki::htmlpage($data->{page});
170                 
171                 $config{aggregateinternal} = 1;
172                 my $newname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page});
173                 
174                 debug "moving $oldname -> $newname";
175                 if (-e $newname) {
176                         if (-e $oldname) {
177                                 error("$newname already exists");
178                         }
179                         else {
180                                 debug("already renamed to $newname?");
181                         }
182                 }
183                 elsif (-e $oldname) {
184                         rename($oldname, $newname) || error("$!");
185                 }
186                 else {
187                         debug("$oldname not found");
188                 }
189                 if (-e $oldoutput) {
190                         require IkiWiki::Render;
191                         debug("removing output file $oldoutput");
192                         IkiWiki::prune($oldoutput);
193                 }
194         }
195         
196         savestate();
197         IkiWiki::unlockwiki;
198         
199         unlockaggregate();
200 }
201
202 sub needsbuild (@) {
203         my $needsbuild=shift;
204         
205         loadstate();
206
207         foreach my $feed (values %feeds) {
208                 if (exists $pagesources{$feed->{sourcepage}} && 
209                     grep { $_ eq $pagesources{$feed->{sourcepage}} } @$needsbuild) {
210                         # Mark all feeds originating on this page as 
211                         # not yet seen; preprocess will unmark those that
212                         # still exist.
213                         markunseen($feed->{sourcepage});
214                 }
215         }
216
217         return $needsbuild;
218 }
219
220 sub preprocess (@) {
221         my %params=@_;
222
223         foreach my $required (qw{name url}) {
224                 if (! exists $params{$required}) {
225                         error sprintf(gettext("missing %s parameter"), $required)
226                 }
227         }
228
229         my $feed={};
230         my $name=$params{name};
231         if (exists $feeds{$name}) {
232                 $feed=$feeds{$name};
233         }
234         else {
235                 $feeds{$name}=$feed;
236         }
237         $feed->{name}=$name;
238         $feed->{sourcepage}=$params{page};
239         $feed->{url}=$params{url};
240         my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".titlepage($params{name});
241         $dir=~s/^\/+//;
242         ($dir)=$dir=~/$config{wiki_file_regexp}/;
243         $feed->{dir}=$dir;
244         $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
245         $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
246         $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
247         $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
248         if (exists $params{template}) {
249                 $params{template}=~s/[^-_a-zA-Z0-9]+//g;
250         }
251         else {
252                 $params{template} = "aggregatepost"
253         }
254         $feed->{template}=$params{template} . ".tmpl";
255         delete $feed->{unseen};
256         $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
257         $feed->{lasttry}=$feed->{lastupdate} unless defined $feed->{lasttry};
258         $feed->{numposts}=0 unless defined $feed->{numposts};
259         $feed->{newposts}=0 unless defined $feed->{newposts};
260         $feed->{message}=gettext("new feed") unless defined $feed->{message};
261         $feed->{error}=0 unless defined $feed->{error};
262         $feed->{tags}=[];
263         while (@_) {
264                 my $key=shift;
265                 my $value=shift;
266                 if ($key eq 'tag') {
267                         push @{$feed->{tags}}, $value;
268                 }
269         }
270
271         return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
272                ($feed->{error} ? "<em>" : "").$feed->{message}.
273                ($feed->{error} ? "</em>" : "").
274                " (".$feed->{numposts}." ".gettext("posts").
275                ($feed->{newposts} ? "; ".$feed->{newposts}.
276                                     " ".gettext("new") : "").
277                ")";
278 }
279
280 sub delete (@) {
281         my @files=@_;
282
283         # Remove feed data for removed pages.
284         foreach my $file (@files) {
285                 my $page=pagename($file);
286                 markunseen($page);
287         }
288 }
289
290 sub markunseen ($) {
291         my $page=shift;
292
293         foreach my $id (keys %feeds) {
294                 if ($feeds{$id}->{sourcepage} eq $page) {
295                         $feeds{$id}->{unseen}=1;
296                 }
297         }
298 }
299
300 my $state_loaded=0;
301
302 sub loadstate () {
303         return if $state_loaded;
304         $state_loaded=1;
305         if (-e "$config{wikistatedir}/aggregate") {
306                 open(IN, "<", "$config{wikistatedir}/aggregate") ||
307                         die "$config{wikistatedir}/aggregate: $!";
308                 while (<IN>) {
309                         $_=IkiWiki::possibly_foolish_untaint($_);
310                         chomp;
311                         my $data={};
312                         foreach my $i (split(/ /, $_)) {
313                                 my ($field, $val)=split(/=/, $i, 2);
314                                 if ($field eq "name" || $field eq "feed" ||
315                                     $field eq "guid" || $field eq "message") {
316                                         $data->{$field}=decode_entities($val, " \t\n");
317                                 }
318                                 elsif ($field eq "tag") {
319                                         push @{$data->{tags}}, $val;
320                                 }
321                                 else {
322                                         $data->{$field}=$val;
323                                 }
324                         }
325                         
326                         if (exists $data->{name}) {
327                                 $feeds{$data->{name}}=$data;
328                         }
329                         elsif (exists $data->{guid}) {
330                                 $guids{$data->{guid}}=$data;
331                         }
332                 }
333
334                 close IN;
335         }
336 }
337
338 sub savestate () {
339         return unless $state_loaded;
340         garbage_collect();
341         my $newfile="$config{wikistatedir}/aggregate.new";
342         my $cleanup = sub { unlink($newfile) };
343         open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup);
344         foreach my $data (values %feeds, values %guids) {
345                 my @line;
346                 foreach my $field (keys %$data) {
347                         if ($field eq "name" || $field eq "feed" ||
348                             $field eq "guid" || $field eq "message") {
349                                 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
350                         }
351                         elsif ($field eq "tags") {
352                                 push @line, "tag=$_" foreach @{$data->{tags}};
353                         }
354                         else {
355                                 push @line, "$field=".$data->{$field}
356                                         if defined $data->{$field};
357                         }
358                 }
359                 print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
360         }
361         close OUT || error("save $newfile: $!", $cleanup);
362         rename($newfile, "$config{wikistatedir}/aggregate") ||
363                 error("rename $newfile: $!", $cleanup);
364
365         my $timestamp=undef;
366         foreach my $feed (keys %feeds) {
367                 my $t=$feeds{$feed}->{lastupdate}+$feeds{$feed}->{updateinterval};
368                 if (! defined $timestamp || $timestamp > $t) {
369                         $timestamp=$t;
370                 }
371         }
372         $newfile=~s/\.new$/time/;
373         open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup);
374         if (defined $timestamp) {
375                 print OUT $timestamp."\n";
376         }
377         close OUT || error("save $newfile: $!", $cleanup);
378 }
379
380 sub garbage_collect () {
381         foreach my $name (keys %feeds) {
382                 # remove any feeds that were not seen while building the pages
383                 # that used to contain them
384                 if ($feeds{$name}->{unseen}) {
385                         delete $feeds{$name};
386                 }
387         }
388
389         foreach my $guid (values %guids) {
390                 # any guid whose feed is gone should be removed
391                 if (! exists $feeds{$guid->{feed}}) {
392                         if (exists $guid->{page}) {
393                                 unlink "$config{srcdir}/".htmlfn($guid->{page});
394                                 unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page});
395                         }
396                         delete $guids{$guid->{guid}};
397                 }
398                 # handle expired guids
399                 elsif ($guid->{expired} && exists $guid->{page}) {
400                         unlink "$config{srcdir}/".htmlfn($guid->{page});
401                         unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page});
402                         delete $guid->{page};
403                         delete $guid->{md5};
404                 }
405         }
406 }
407
408 sub mergestate () {
409         # Load the current state in from disk, and merge into it
410         # values from the state in memory that might have changed
411         # during aggregation.
412         my %myfeeds=%feeds;
413         my %myguids=%guids;
414         clearstate();
415         loadstate();
416
417         # All that can change in feed state during aggregation is a few
418         # fields.
419         foreach my $name (keys %myfeeds) {
420                 if (exists $feeds{$name}) {
421                         foreach my $field (qw{message lastupdate lasttry
422                                               numposts newposts error}) {
423                                 $feeds{$name}->{$field}=$myfeeds{$name}->{$field};
424                         }
425                 }
426         }
427
428         # New guids can be created during aggregation.
429         # Guids have a few fields that may be updated during aggregation.
430         # It's also possible that guids were removed from the on-disk state
431         # while the aggregation was in process. That would only happen if
432         # their feed was also removed, so any removed guids added back here
433         # will be garbage collected later.
434         foreach my $guid (keys %myguids) {
435                 if (! exists $guids{$guid}) {
436                         $guids{$guid}=$myguids{$guid};
437                 }
438                 else {
439                         foreach my $field (qw{md5}) {
440                                 $guids{$guid}->{$field}=$myguids{$guid}->{$field};
441                         }
442                 }
443         }
444 }
445
446 sub clearstate () {
447         %feeds=();
448         %guids=();
449         $state_loaded=0;
450 }
451
452 sub expire () {
453         foreach my $feed (values %feeds) {
454                 next unless $feed->{expireage} || $feed->{expirecount};
455                 my $count=0;
456                 my %seen;
457                 foreach my $item (sort { ($IkiWiki::pagectime{$b->{page}} || 0) <=> ($IkiWiki::pagectime{$a->{page}} || 0) }
458                                   grep { exists $_->{page} && $_->{feed} eq $feed->{name} }
459                                   values %guids) {
460                         if ($feed->{expireage}) {
461                                 my $days_old = (time - ($IkiWiki::pagectime{$item->{page}} || 0)) / 60 / 60 / 24;
462                                 if ($days_old > $feed->{expireage}) {
463                                         debug(sprintf(gettext("expiring %s (%s days old)"),
464                                                 $item->{page}, int($days_old)));
465                                         $item->{expired}=1;
466                                 }
467                         }
468                         elsif ($feed->{expirecount} &&
469                                $count >= $feed->{expirecount}) {
470                                 debug(sprintf(gettext("expiring %s"), $item->{page}));
471                                 $item->{expired}=1;
472                         }
473                         else {
474                                 if (! $seen{$item->{page}}) {
475                                         $seen{$item->{page}}=1;
476                                         $count++;
477                                 }
478                         }
479                 }
480         }
481 }
482
483 sub needsaggregate () {
484         return values %feeds if $config{rebuild};
485         return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds;
486 }
487
488 sub aggregate (@) {
489         eval q{use XML::Feed};
490         error($@) if $@;
491         eval q{use URI::Fetch};
492         error($@) if $@;
493
494         foreach my $feed (@_) {
495                 $feed->{lasttry}=time;
496                 $feed->{newposts}=0;
497                 $feed->{message}=sprintf(gettext("last checked %s"),
498                         displaytime($feed->{lasttry}));
499                 $feed->{error}=0;
500
501                 debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
502
503                 if (! length $feed->{feedurl}) {
504                         my @urls=XML::Feed->find_feeds($feed->{url});
505                         if (! @urls) {
506                                 $feed->{message}=sprintf(gettext("could not find feed at %s"), $feed->{url});
507                                 $feed->{error}=1;
508                                 debug($feed->{message});
509                                 next;
510                         }
511                         $feed->{feedurl}=pop @urls;
512                 }
513                 my $res=URI::Fetch->fetch($feed->{feedurl});
514                 if (! $res) {
515                         $feed->{message}=URI::Fetch->errstr;
516                         $feed->{error}=1;
517                         debug($feed->{message});
518                         next;
519                 }
520
521                 # lastupdate is only set if we were able to contact the server
522                 $feed->{lastupdate}=$feed->{lasttry};
523
524                 if ($res->status == URI::Fetch::URI_GONE()) {
525                         $feed->{message}=gettext("feed not found");
526                         $feed->{error}=1;
527                         debug($feed->{message});
528                         next;
529                 }
530                 my $content=$res->content;
531                 my $f=eval{XML::Feed->parse(\$content)};
532                 if ($@) {
533                         # One common cause of XML::Feed crashing is a feed
534                         # that contains invalid UTF-8 sequences. Convert
535                         # feed to ascii to try to work around.
536                         $feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)"));
537                         $f=eval {
538                                 $content=Encode::decode_utf8($content, 0);
539                                 XML::Feed->parse(\$content)
540                         };
541                 }
542                 if ($@) {
543                         # Another possibility is badly escaped entities.
544                         $feed->{message}.=" ".sprintf(gettext("(feed entities escaped)"));
545                         $content=~s/\&(?!amp)(\w+);/&amp;$1;/g;
546                         $f=eval {
547                                 $content=Encode::decode_utf8($content, 0);
548                                 XML::Feed->parse(\$content)
549                         };
550                 }
551                 if ($@) {
552                         $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
553                         $feed->{error}=1;
554                         debug($feed->{message});
555                         next;
556                 }
557                 if (! $f) {
558                         $feed->{message}=XML::Feed->errstr;
559                         $feed->{error}=1;
560                         debug($feed->{message});
561                         next;
562                 }
563
564                 foreach my $entry ($f->entries) {
565                         # XML::Feed doesn't work around XML::Atom's bizarre
566                         # API, so we will. Real unicode strings? Yes please.
567                         # See [[bugs/Aggregated_Atom_feeds_are_double-encoded]]
568                         local $XML::Atom::ForceUnicode = 1;
569
570                         my $c=$entry->content;
571                         # atom feeds may have no content, only a summary
572                         if (! defined $c && ref $entry->summary) {
573                                 $c=$entry->summary;
574                         }
575
576                         add_page(
577                                 feed => $feed,
578                                 copyright => $f->copyright,
579                                 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
580                                 link => $entry->link,
581                                 content => (defined $c && defined $c->body) ? $c->body : "",
582                                 guid => defined $entry->id ? $entry->id : time."_".$feed->{name},
583                                 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
584                                 base => (defined $c && $c->can("base")) ? $c->base : undef,
585                         );
586                 }
587         }
588 }
589
590 sub add_page (@) {
591         my %params=@_;
592         
593         my $feed=$params{feed};
594         my $guid={};
595         my $mtime;
596         if (exists $guids{$params{guid}}) {
597                 # updating an existing post
598                 $guid=$guids{$params{guid}};
599                 return if $guid->{expired};
600         }
601         else {
602                 # new post
603                 $guid->{guid}=$params{guid};
604                 $guids{$params{guid}}=$guid;
605                 $mtime=$params{ctime};
606                 $feed->{numposts}++;
607                 $feed->{newposts}++;
608
609                 # assign it an unused page
610                 my $page=titlepage($params{title});
611                 # escape slashes and periods in title so it doesn't specify
612                 # directory name or trigger ".." disallowing code.
613                 $page=~s!([/.])!"__".ord($1)."__"!eg;
614                 $page=$feed->{dir}."/".$page;
615                 ($page)=$page=~/$config{wiki_file_regexp}/;
616                 if (! defined $page || ! length $page) {
617                         $page=$feed->{dir}."/item";
618                 }
619                 my $c="";
620                 while (exists $IkiWiki::pagecase{lc $page.$c} ||
621                        -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) ||
622                        -e "$config{srcdir}/".htmlfn($page.$c)) {
623                         $c++
624                 }
625
626                 # Make sure that the file name isn't too long. 
627                 # NB: This doesn't check for path length limits.
628                 my $max=POSIX::pathconf($config{srcdir}, &POSIX::_PC_NAME_MAX);
629                 if (defined $max && length(htmlfn($page)) >= $max) {
630                         $c="";
631                         $page=$feed->{dir}."/item";
632                         while (exists $IkiWiki::pagecase{lc $page.$c} ||
633                               -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) ||
634
635                                -e "$config{srcdir}/".htmlfn($page.$c)) {
636                                 $c++
637                         }
638                 }
639
640                 $guid->{page}=$page;
641                 debug(sprintf(gettext("creating new page %s"), $page));
642         }
643         $guid->{feed}=$feed->{name};
644         
645         # To write or not to write? Need to avoid writing unchanged pages
646         # to avoid unneccessary rebuilding. The mtime from rss cannot be
647         # trusted; let's use a digest.
648         eval q{use Digest::MD5 'md5_hex'};
649         error($@) if $@;
650         require Encode;
651         my $digest=md5_hex(Encode::encode_utf8($params{content}));
652         return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
653         $guid->{md5}=$digest;
654
655         # Create the page.
656         my $template;
657         eval {
658                 $template=template($feed->{template}, blind_cache => 1);
659         };
660         if ($@) {
661                 print STDERR gettext("failed to process template:")." $@";
662                 return;
663         }
664         $template->param(title => $params{title})
665                 if defined $params{title} && length($params{title});
666         $template->param(content => wikiescape(htmlabs($params{content},
667                 defined $params{base} ? $params{base} : $feed->{feedurl})));
668         $template->param(name => $feed->{name});
669         $template->param(url => $feed->{url});
670         $template->param(copyright => $params{copyright})
671                 if defined $params{copyright} && length $params{copyright};
672         $template->param(permalink => IkiWiki::urlabs($params{link}, $feed->{feedurl}))
673                 if defined $params{link};
674         if (ref $feed->{tags}) {
675                 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
676         }
677         writefile(htmlfn($guid->{page}),
678                 $IkiWiki::Plugin::transient::transientdir, $template->output);
679
680         if (defined $mtime && $mtime <= time) {
681                 # Set the mtime, this lets the build process get the right
682                 # creation time on record for the new page.
683                 utime $mtime, $mtime,
684                         $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page});
685                 # Store it in pagectime for expiry code to use also.
686                 $IkiWiki::pagectime{$guid->{page}}=$mtime
687                         unless exists $IkiWiki::pagectime{$guid->{page}};
688         }
689         else {
690                 # Dummy value for expiry code.
691                 $IkiWiki::pagectime{$guid->{page}}=time
692                         unless exists $IkiWiki::pagectime{$guid->{page}};
693         }
694 }
695
696 sub wikiescape ($) {
697         # escape accidental wikilinks and preprocessor stuff
698         return encode_entities(shift, '\[\]');
699 }
700
701 sub htmlabs ($$) {
702         # Convert links in html from relative to absolute.
703         # Note that this is a heuristic, which is not specified by the rss
704         # spec and may not be right for all feeds. Also, see Debian
705         # bug #381359.
706         my $html=shift;
707         my $urlbase=shift;
708
709         my $ret="";
710         my $p = HTML::Parser->new(api_version => 3);
711         $p->handler(default => sub { $ret.=join("", @_) }, "text");
712         $p->handler(start => sub {
713                 my ($tagname, $pos, $text) = @_;
714                 if (ref $HTML::Tagset::linkElements{$tagname}) {
715                         while (4 <= @$pos) {
716                                 # use attribute sets from right to left
717                                 # to avoid invalidating the offsets
718                                 # when replacing the values
719                                 my($k_offset, $k_len, $v_offset, $v_len) =
720                                         splice(@$pos, -4);
721                                 my $attrname = lc(substr($text, $k_offset, $k_len));
722                                 next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
723                                 next unless $v_offset; # 0 v_offset means no value
724                                 my $v = substr($text, $v_offset, $v_len);
725                                 $v =~ s/^([\'\"])(.*)\1$/$2/;
726                                 my $new_v=IkiWiki::urlabs($v, $urlbase);
727                                 $new_v =~ s/\"/&quot;/g; # since we quote with ""
728                                 substr($text, $v_offset, $v_len) = qq("$new_v");
729                         }
730                 }
731                 $ret.=$text;
732         }, "tagname, tokenpos, text");
733         $p->parse($html);
734         $p->eof;
735
736         return $ret;
737 }
738
739 sub htmlfn ($) {
740         return shift().".".($config{aggregateinternal} ? "_aggregated" : $config{htmlext});
741 }
742
743 my $aggregatelock;
744
745 sub lockaggregate () {
746         # Take an exclusive lock to prevent multiple concurrent aggregators.
747         # Returns true if the lock was aquired.
748         if (! -d $config{wikistatedir}) {
749                 mkdir($config{wikistatedir});
750         }
751         open($aggregatelock, '>', "$config{wikistatedir}/aggregatelock") ||
752                 error ("cannot open to $config{wikistatedir}/aggregatelock: $!");
753         if (! flock($aggregatelock, 2 | 4)) { # LOCK_EX | LOCK_NB
754                 close($aggregatelock) || error("failed closing aggregatelock: $!");
755                 return 0;
756         }
757         return 1;
758 }
759
760 sub unlockaggregate () {
761         return close($aggregatelock) if $aggregatelock;
762         return;
763 }
764
765 1