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