git.oblomov.eu Git - ikiwiki/blob - 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                 last => 1);
  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 $@;
  34         Getopt::Long::Configure('pass_through');
  35         GetOptions(
  36                 "aggregate" => \$config{aggregate},
  37                 "aggregateinternal!" => \$config{aggregateinternal},
  38         );
  39 }
  40
  41 sub getsetup () {
  42         return
  43                 plugin => {
  44                         safe => 1,
  45                         rebuild => undef,
  46                 },
  47                 aggregateinternal => {
  48                         type => "boolean",
  49                         example => 1,
  50                         description => "enable aggregation to internal pages?",
  51                         safe => 0, # enabling needs manual transition
  52                         rebuild => 0,
  53                 },
  54                 aggregate_webtrigger => {
  55                         type => "boolean",
  56                         example => 0,
  57                         description => "allow aggregation to be triggered via the web?",
  58                         safe => 1,
  59                         rebuild => 0,
  60                 },
  61                 cookiejar => {
  62                         type => "string",
  63                         example => { file => "$ENV{HOME}/.ikiwiki/cookies" },
  64                         safe => 0, # hooks into perl module internals
  65                         description => "cookie control",
  66                 },
  67 }
  68
  69 sub checkconfig () {
  70         if (! defined $config{aggregateinternal}) {
  71                 $config{aggregateinternal}=1;
  72         }
  73         if (! defined $config{cookies}) {
  74                 $config{cookies}={ file => "$ENV{HOME}/.ikiwiki/cookies" };
  75         }
  76
  77         if ($config{aggregate} && ! ($config{post_commit} &&
  78                                      IkiWiki::commit_hook_enabled())) {
  79                 launchaggregation();
  80         }
  81 }
  82
  83 sub cgi ($) {
  84         my $cgi=shift;
  85
  86         if (defined $cgi->param('do') &&
  87             $cgi->param("do") eq "aggregate_webtrigger") {
  88                 $|=1;
  89                 print "Content-Type: text/plain\n\n";
  90                 $config{cgi}=0;
  91                 $config{verbose}=1;
  92                 $config{syslog}=0;
  93                 print gettext("Aggregation triggered via web.")."\n\n";
  94                 if (launchaggregation()) {
  95                         IkiWiki::lockwiki();
  96                         IkiWiki::loadindex();
  97                         require IkiWiki::Render;
  98                         IkiWiki::refresh();
  99                         IkiWiki::saveindex();
 100                 }
 101                 else {
 102                         print gettext("Nothing to do right now, all feeds are up-to-date!")."\n";
 103                 }
 104                 exit 0;
 105         }
 106 }
 107
 108 sub launchaggregation () {
 109         # See if any feeds need aggregation.
 110         loadstate();
 111         my @feeds=needsaggregate();
 112         return unless @feeds;
 113         if (! lockaggregate()) {
 114                 debug("an aggregation process is already running");
 115                 return;
 116         }
 117         # force a later rebuild of source pages
 118         $IkiWiki::forcerebuild{$_->{sourcepage}}=1
 119                 foreach @feeds;
 120
 121         # Fork a child process to handle the aggregation.
 122         # The parent process will then handle building the
 123         # result. This avoids messy code to clear state
 124         # accumulated while aggregating.
 125         defined(my $pid = fork) or error("Can't fork: $!");
 126         if (! $pid) {
 127                 IkiWiki::loadindex();
 128                 # Aggregation happens without the main wiki lock
 129                 # being held. This allows editing pages etc while
 130                 # aggregation is running.
 131                 aggregate(@feeds);
 132
 133                 IkiWiki::lockwiki;
 134                 # Merge changes, since aggregation state may have
 135                 # changed on disk while the aggregation was happening.
 136                 mergestate();
 137                 expire();
 138                 savestate();
 139                 IkiWiki::unlockwiki;
 140                 exit 0;
 141         }
 142         waitpid($pid,0);
 143         if ($?) {
 144                 error "aggregation failed with code $?";
 145         }
 146
 147         clearstate();
 148         unlockaggregate();
 149
 150         return 1;
 151 }
 152
 153 #  Pages with extension _aggregated have plain html markup, pass through.
 154 sub htmlize (@) {
 155         my %params=@_;
 156         return $params{content};
 157 }
 158
 159 # Used by ikiwiki-transition aggregateinternal.
 160 sub migrate_to_internal {
 161         if (! lockaggregate()) {
 162                 error("an aggregation process is currently running");
 163         }
 164
 165         IkiWiki::lockwiki();
 166         loadstate();
 167         $config{verbose}=1;
 168
 169         foreach my $data (values %guids) {
 170                 next unless $data->{page};
 171                 next if $data->{expired};
 172
 173                 $config{aggregateinternal} = 0;
 174                 my $oldname = "$config{srcdir}/".htmlfn($data->{page});
 175                 if (! -e $oldname) {
 176                         $oldname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page});
 177                 }
 178
 179                 my $oldoutput = $config{destdir}."/".IkiWiki::htmlpage($data->{page});
 180
 181                 $config{aggregateinternal} = 1;
 182                 my $newname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page});
 183
 184                 debug "moving $oldname -> $newname";
 185                 if (-e $newname) {
 186                         if (-e $oldname) {
 187                                 error("$newname already exists");
 188                         }
 189                         else {
 190                                 debug("already renamed to $newname?");
 191                         }
 192                 }
 193                 elsif (-e $oldname) {
 194                         rename($oldname, $newname) || error("$!");
 195                 }
 196                 else {
 197                         debug("$oldname not found");
 198                 }
 199                 if (-e $oldoutput) {
 200                         require IkiWiki::Render;
 201                         debug("removing output file $oldoutput");
 202                         IkiWiki::prune($oldoutput);
 203                 }
 204         }
 205
 206         savestate();
 207         IkiWiki::unlockwiki;
 208
 209         unlockaggregate();
 210 }
 211
 212 sub needsbuild (@) {
 213         my $needsbuild=shift;
 214
 215         loadstate();
 216
 217         foreach my $feed (values %feeds) {
 218                 if (exists $pagesources{$feed->{sourcepage}} &&
 219                     grep { $_ eq $pagesources{$feed->{sourcepage}} } @$needsbuild) {
 220                         # Mark all feeds originating on this page as
 221                         # not yet seen; preprocess will unmark those that
 222                         # still exist.
 223                         markunseen($feed->{sourcepage});
 224                 }
 225         }
 226
 227         return $needsbuild;
 228 }
 229
 230 sub preprocess (@) {
 231         my %params=@_;
 232
 233         foreach my $required (qw{name url}) {
 234                 if (! exists $params{$required}) {
 235                         error sprintf(gettext("missing %s parameter"), $required)
 236                 }
 237         }
 238
 239         my $feed={};
 240         my $name=$params{name};
 241         if (exists $feeds{$name}) {
 242                 $feed=$feeds{$name};
 243         }
 244         else {
 245                 $feeds{$name}=$feed;
 246         }
 247         $feed->{name}=$name;
 248         $feed->{sourcepage}=$params{page};
 249         $feed->{url}=$params{url};
 250         my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".titlepage($params{name});
 251         $dir=~s/^\/+//;
 252         ($dir)=$dir=~/$config{wiki_file_regexp}/;
 253         $feed->{dir}=$dir;
 254         $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
 255         $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
 256         $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
 257         $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
 258         if (exists $params{template}) {
 259                 $params{template}=~s/[^-_a-zA-Z0-9]+//g;
 260         }
 261         else {
 262                 $params{template} = "aggregatepost"
 263         }
 264         $feed->{template}=$params{template} . ".tmpl";
 265         delete $feed->{unseen};
 266         $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
 267         $feed->{lasttry}=$feed->{lastupdate} unless defined $feed->{lasttry};
 268         $feed->{numposts}=0 unless defined $feed->{numposts};
 269         $feed->{newposts}=0 unless defined $feed->{newposts};
 270         $feed->{message}=gettext("new feed") unless defined $feed->{message};
 271         $feed->{error}=0 unless defined $feed->{error};
 272         $feed->{tags}=[];
 273         while (@_) {
 274                 my $key=shift;
 275                 my $value=shift;
 276                 if ($key eq 'tag') {
 277                         push @{$feed->{tags}}, $value;
 278                 }
 279         }
 280
 281         return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
 282                ($feed->{error} ? "<em>" : "").$feed->{message}.
 283                ($feed->{error} ? "</em>" : "").
 284                " (".$feed->{numposts}." ".gettext("posts").
 285                ($feed->{newposts} ? "; ".$feed->{newposts}.
 286                                     " ".gettext("new") : "").
 287                ")";
 288 }
 289
 290 sub delete (@) {
 291         my @files=@_;
 292
 293         # Remove feed data for removed pages.
 294         foreach my $file (@files) {
 295                 my $page=pagename($file);
 296                 markunseen($page);
 297         }
 298 }
 299
 300 sub markunseen ($) {
 301         my $page=shift;
 302
 303         foreach my $id (keys %feeds) {
 304                 if ($feeds{$id}->{sourcepage} eq $page) {
 305                         $feeds{$id}->{unseen}=1;
 306                 }
 307         }
 308 }
 309
 310 my $state_loaded=0;
 311
 312 sub loadstate () {
 313         return if $state_loaded;
 314         $state_loaded=1;
 315         if (-e "$config{wikistatedir}/aggregate") {
 316                 open(IN, "<", "$config{wikistatedir}/aggregate") ||
 317                         die "$config{wikistatedir}/aggregate: $!";
 318                 while (<IN>) {
 319                         $_=IkiWiki::possibly_foolish_untaint($_);
 320                         chomp;
 321                         my $data={};
 322                         foreach my $i (split(/ /, $_)) {
 323                                 my ($field, $val)=split(/=/, $i, 2);
 324                                 if ($field eq "name" || $field eq "feed" ||
 325                                     $field eq "guid" || $field eq "message") {
 326                                         $data->{$field}=decode_entities($val, " \t\n");
 327                                 }
 328                                 elsif ($field eq "tag") {
 329                                         push @{$data->{tags}}, $val;
 330                                 }
 331                                 else {
 332                                         $data->{$field}=$val;
 333                                 }
 334                         }
 335
 336                         if (exists $data->{name}) {
 337                                 $feeds{$data->{name}}=$data;
 338                         }
 339                         elsif (exists $data->{guid}) {
 340                                 $guids{$data->{guid}}=$data;
 341                         }
 342                 }
 343
 344                 close IN;
 345         }
 346 }
 347
 348 sub savestate () {
 349         return unless $state_loaded;
 350         garbage_collect();
 351         my $newfile="$config{wikistatedir}/aggregate.new";
 352         my $cleanup = sub { unlink($newfile) };
 353         open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup);
 354         foreach my $data (values %feeds, values %guids) {
 355                 my @line;
 356                 foreach my $field (keys %$data) {
 357                         if ($field eq "name" || $field eq "feed" ||
 358                             $field eq "guid" || $field eq "message") {
 359                                 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
 360                         }
 361                         elsif ($field eq "tags") {
 362                                 push @line, "tag=$_" foreach @{$data->{tags}};
 363                         }
 364                         else {
 365                                 push @line, "$field=".$data->{$field}
 366                                         if defined $data->{$field};
 367                         }
 368                 }
 369                 print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
 370         }
 371         close OUT || error("save $newfile: $!", $cleanup);
 372         rename($newfile, "$config{wikistatedir}/aggregate") ||
 373                 error("rename $newfile: $!", $cleanup);
 374
 375         my $timestamp=undef;
 376         foreach my $feed (keys %feeds) {
 377                 my $t=$feeds{$feed}->{lastupdate}+$feeds{$feed}->{updateinterval};
 378                 if (! defined $timestamp || $timestamp > $t) {
 379                         $timestamp=$t;
 380                 }
 381         }
 382         $newfile=~s/\.new$/time/;
 383         open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup);
 384         if (defined $timestamp) {
 385                 print OUT $timestamp."\n";
 386         }
 387         close OUT || error("save $newfile: $!", $cleanup);
 388 }
 389
 390 sub garbage_collect () {
 391         foreach my $name (keys %feeds) {
 392                 # remove any feeds that were not seen while building the pages
 393                 # that used to contain them
 394                 if ($feeds{$name}->{unseen}) {
 395                         delete $feeds{$name};
 396                 }
 397         }
 398
 399         foreach my $guid (values %guids) {
 400                 # any guid whose feed is gone should be removed
 401                 if (! exists $feeds{$guid->{feed}}) {
 402                         if (exists $guid->{page}) {
 403                                 unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page})
 404                                         || unlink "$config{srcdir}/".htmlfn($guid->{page});
 405                         }
 406                         delete $guids{$guid->{guid}};
 407                 }
 408                 # handle expired guids
 409                 elsif ($guid->{expired} && exists $guid->{page}) {
 410                         unlink "$config{srcdir}/".htmlfn($guid->{page});
 411                         unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page});
 412                         delete $guid->{page};
 413                         delete $guid->{md5};
 414                 }
 415         }
 416 }
 417
 418 sub mergestate () {
 419         # Load the current state in from disk, and merge into it
 420         # values from the state in memory that might have changed
 421         # during aggregation.
 422         my %myfeeds=%feeds;
 423         my %myguids=%guids;
 424         clearstate();
 425         loadstate();
 426
 427         # All that can change in feed state during aggregation is a few
 428         # fields.
 429         foreach my $name (keys %myfeeds) {
 430                 if (exists $feeds{$name}) {
 431                         foreach my $field (qw{message lastupdate lasttry
 432                                               numposts newposts error}) {
 433                                 $feeds{$name}->{$field}=$myfeeds{$name}->{$field};
 434                         }
 435                 }
 436         }
 437
 438         # New guids can be created during aggregation.
 439         # Guids have a few fields that may be updated during aggregation.
 440         # It's also possible that guids were removed from the on-disk state
 441         # while the aggregation was in process. That would only happen if
 442         # their feed was also removed, so any removed guids added back here
 443         # will be garbage collected later.
 444         foreach my $guid (keys %myguids) {
 445                 if (! exists $guids{$guid}) {
 446                         $guids{$guid}=$myguids{$guid};
 447                 }
 448                 else {
 449                         foreach my $field (qw{md5}) {
 450                                 $guids{$guid}->{$field}=$myguids{$guid}->{$field};
 451                         }
 452                 }
 453         }
 454 }
 455
 456 sub clearstate () {
 457         %feeds=();
 458         %guids=();
 459         $state_loaded=0;
 460 }
 461
 462 sub expire () {
 463         foreach my $feed (values %feeds) {
 464                 next unless $feed->{expireage} || $feed->{expirecount};
 465                 my $count=0;
 466                 my %seen;
 467                 foreach my $item (sort { ($IkiWiki::pagectime{$b->{page}} || 0) <=> ($IkiWiki::pagectime{$a->{page}} || 0) }
 468                                   grep { exists $_->{page} && $_->{feed} eq $feed->{name} }
 469                                   values %guids) {
 470                         if ($feed->{expireage}) {
 471                                 my $days_old = (time - ($IkiWiki::pagectime{$item->{page}} || 0)) / 60 / 60 / 24;
 472                                 if ($days_old > $feed->{expireage}) {
 473                                         debug(sprintf(gettext("expiring %s (%s days old)"),
 474                                                 $item->{page}, int($days_old)));
 475                                         $item->{expired}=1;
 476                                 }
 477                         }
 478                         elsif ($feed->{expirecount} &&
 479                                $count >= $feed->{expirecount}) {
 480                                 debug(sprintf(gettext("expiring %s"), $item->{page}));
 481                                 $item->{expired}=1;
 482                         }
 483                         else {
 484                                 if (! $seen{$item->{page}}) {
 485                                         $seen{$item->{page}}=1;
 486                                         $count++;
 487                                 }
 488                         }
 489                 }
 490         }
 491 }
 492
 493 sub needsaggregate () {
 494         return values %feeds if $config{rebuild};
 495         return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds;
 496 }
 497
 498 sub aggregate (@) {
 499         eval q{use XML::Feed};
 500         error($@) if $@;
 501         eval q{use URI::Fetch};
 502         error($@) if $@;
 503
 504         foreach my $feed (@_) {
 505                 $feed->{lasttry}=time;
 506                 $feed->{newposts}=0;
 507                 $feed->{message}=sprintf(gettext("last checked %s"),
 508                         displaytime($feed->{lasttry}));
 509                 $feed->{error}=0;
 510
 511                 debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
 512
 513                 if (! length $feed->{feedurl}) {
 514                         my @urls=XML::Feed->find_feeds($feed->{url});
 515                         if (! @urls) {
 516                                 $feed->{message}=sprintf(gettext("could not find feed at %s"), $feed->{url});
 517                                 $feed->{error}=1;
 518                                 debug($feed->{message});
 519                                 next;
 520                         }
 521                         $feed->{feedurl}=pop @urls;
 522                 }
 523                 my $res=URI::Fetch->fetch($feed->{feedurl},
 524                         UserAgent => LWP::UserAgent->new(
 525                                 cookie_jar => $config{cookiejar},
 526                         ),
 527                 );
 528                 if (! $res) {
 529                         $feed->{message}=URI::Fetch->errstr;
 530                         $feed->{error}=1;
 531                         debug($feed->{message});
 532                         next;
 533                 }
 534
 535                 # lastupdate is only set if we were able to contact the server
 536                 $feed->{lastupdate}=$feed->{lasttry};
 537
 538                 if ($res->status == URI::Fetch::URI_GONE()) {
 539                         $feed->{message}=gettext("feed not found");
 540                         $feed->{error}=1;
 541                         debug($feed->{message});
 542                         next;
 543                 }
 544                 my $content=$res->content;
 545                 my $f=eval{XML::Feed->parse(\$content)};
 546                 if ($@) {
 547                         # One common cause of XML::Feed crashing is a feed
 548                         # that contains invalid UTF-8 sequences. Convert
 549                         # feed to ascii to try to work around.
 550                         $feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)"));
 551                         $f=eval {
 552                                 $content=Encode::decode_utf8($content, 0);
 553                                 XML::Feed->parse(\$content)
 554                         };
 555                 }
 556                 if ($@) {
 557                         # Another possibility is badly escaped entities.
 558                         $feed->{message}.=" ".sprintf(gettext("(feed entities escaped)"));
 559                         $content=~s/\&(?!amp)(\w+);/&amp;$1;/g;
 560                         $f=eval {
 561                                 $content=Encode::decode_utf8($content, 0);
 562                                 XML::Feed->parse(\$content)
 563                         };
 564                 }
 565                 if ($@) {
 566                         $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
 567                         $feed->{error}=1;
 568                         debug($feed->{message});
 569                         next;
 570                 }
 571                 if (! $f) {
 572                         $feed->{message}=XML::Feed->errstr;
 573                         $feed->{error}=1;
 574                         debug($feed->{message});
 575                         next;
 576                 }
 577
 578                 foreach my $entry ($f->entries) {
 579                         # XML::Feed doesn't work around XML::Atom's bizarre
 580                         # API, so we will. Real unicode strings? Yes please.
 581                         # See [[bugs/Aggregated_Atom_feeds_are_double-encoded]]
 582                         local $XML::Atom::ForceUnicode = 1;
 583
 584                         my $c=$entry->content;
 585                         # atom feeds may have no content, only a summary
 586                         if (! defined $c && ref $entry->summary) {
 587                                 $c=$entry->summary;
 588                         }
 589
 590                         add_page(
 591                                 feed => $feed,
 592                                 copyright => $f->copyright,
 593                                 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
 594                                 link => $entry->link,
 595                                 content => (defined $c && defined $c->body) ? $c->body : "",
 596                                 guid => defined $entry->id ? $entry->id : time."_".$feed->{name},
 597                                 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
 598                                 base => (defined $c && $c->can("base")) ? $c->base : undef,
 599                         );
 600                 }
 601         }
 602 }
 603
 604 sub add_page (@) {
 605         my %params=@_;
 606
 607         my $feed=$params{feed};
 608         my $guid={};
 609         my $mtime;
 610         if (exists $guids{$params{guid}}) {
 611                 # updating an existing post
 612                 $guid=$guids{$params{guid}};
 613                 return if $guid->{expired};
 614         }
 615         else {
 616                 # new post
 617                 $guid->{guid}=$params{guid};
 618                 $guids{$params{guid}}=$guid;
 619                 $mtime=$params{ctime};
 620                 $feed->{numposts}++;
 621                 $feed->{newposts}++;
 622
 623                 # assign it an unused page
 624                 my $page=titlepage($params{title});
 625                 # escape slashes and periods in title so it doesn't specify
 626                 # directory name or trigger ".." disallowing code.
 627                 $page=~s!([/.])!"__".ord($1)."__"!eg;
 628                 $page=$feed->{dir}."/".$page;
 629                 ($page)=$page=~/$config{wiki_file_regexp}/;
 630                 if (! defined $page || ! length $page) {
 631                         $page=$feed->{dir}."/item";
 632                 }
 633                 my $c="";
 634                 while (exists $IkiWiki::pagecase{lc $page.$c} ||
 635                        -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) ||
 636                        -e "$config{srcdir}/".htmlfn($page.$c)) {
 637                         $c++
 638                 }
 639
 640                 # Make sure that the file name isn't too long.
 641                 # NB: This doesn't check for path length limits.
 642                 my $max=POSIX::pathconf($config{srcdir}, &POSIX::_PC_NAME_MAX);
 643                 if (defined $max && length(htmlfn($page)) >= $max) {
 644                         $c="";
 645                         $page=$feed->{dir}."/item";
 646                         while (exists $IkiWiki::pagecase{lc $page.$c} ||
 647                               -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) ||
 648
 649                                -e "$config{srcdir}/".htmlfn($page.$c)) {
 650                                 $c++
 651                         }
 652                 }
 653
 654                 $guid->{page}=$page;
 655                 debug(sprintf(gettext("creating new page %s"), $page));
 656         }
 657         $guid->{feed}=$feed->{name};
 658
 659         # To write or not to write? Need to avoid writing unchanged pages
 660         # to avoid unneccessary rebuilding. The mtime from rss cannot be
 661         # trusted; let's use a digest.
 662         eval q{use Digest::MD5 'md5_hex'};
 663         error($@) if $@;
 664         require Encode;
 665         my $digest=md5_hex(Encode::encode_utf8($params{content}));
 666         return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
 667         $guid->{md5}=$digest;
 668
 669         # Create the page.
 670         my $template;
 671         eval {
 672                 $template=template($feed->{template}, blind_cache => 1);
 673         };
 674         if ($@) {
 675                 print STDERR gettext("failed to process template:")." $@";
 676                 return;
 677         }
 678         $template->param(title => $params{title})
 679                 if defined $params{title} && length($params{title});
 680         $template->param(content => wikiescape(htmlabs($params{content},
 681                 defined $params{base} ? $params{base} : $feed->{feedurl})));
 682         $template->param(name => $feed->{name});
 683         $template->param(url => $feed->{url});
 684         $template->param(copyright => $params{copyright})
 685                 if defined $params{copyright} && length $params{copyright};
 686         $template->param(permalink => IkiWiki::urlabs($params{link}, $feed->{feedurl}))
 687                 if defined $params{link};
 688         if (ref $feed->{tags}) {
 689                 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
 690         }
 691         writefile(htmlfn($guid->{page}),
 692                 $IkiWiki::Plugin::transient::transientdir, $template->output);
 693
 694         if (defined $mtime && $mtime <= time) {
 695                 # Set the mtime, this lets the build process get the right
 696                 # creation time on record for the new page.
 697                 utime $mtime, $mtime,
 698                         $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page});
 699                 # Store it in pagectime for expiry code to use also.
 700                 $IkiWiki::pagectime{$guid->{page}}=$mtime
 701                         unless exists $IkiWiki::pagectime{$guid->{page}};
 702         }
 703         else {
 704                 # Dummy value for expiry code.
 705                 $IkiWiki::pagectime{$guid->{page}}=time
 706                         unless exists $IkiWiki::pagectime{$guid->{page}};
 707         }
 708 }
 709
 710 sub wikiescape ($) {
 711         # escape accidental wikilinks and preprocessor stuff
 712         return encode_entities(shift, '\[\]');
 713 }
 714
 715 sub htmlabs ($$) {
 716         # Convert links in html from relative to absolute.
 717         # Note that this is a heuristic, which is not specified by the rss
 718         # spec and may not be right for all feeds. Also, see Debian
 719         # bug #381359.
 720         my $html=shift;
 721         my $urlbase=shift;
 722
 723         my $ret="";
 724         my $p = HTML::Parser->new(api_version => 3);
 725         $p->handler(default => sub { $ret.=join("", @_) }, "text");
 726         $p->handler(start => sub {
 727                 my ($tagname, $pos, $text) = @_;
 728                 if (ref $HTML::Tagset::linkElements{$tagname}) {
 729                         while (4 <= @$pos) {
 730                                 # use attribute sets from right to left
 731                                 # to avoid invalidating the offsets
 732                                 # when replacing the values
 733                                 my($k_offset, $k_len, $v_offset, $v_len) =
 734                                         splice(@$pos, -4);
 735                                 my $attrname = lc(substr($text, $k_offset, $k_len));
 736                                 next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
 737                                 next unless $v_offset; # 0 v_offset means no value
 738                                 my $v = substr($text, $v_offset, $v_len);
 739                                 $v =~ s/^([\'\"])(.*)\1$/$2/;
 740                                 my $new_v=IkiWiki::urlabs($v, $urlbase);
 741                                 $new_v =~ s/\"/&quot;/g; # since we quote with ""
 742                                 substr($text, $v_offset, $v_len) = qq("$new_v");
 743                         }
 744                 }
 745                 $ret.=$text;
 746         }, "tagname, tokenpos, text");
 747         $p->parse($html);
 748         $p->eof;
 749
 750         return $ret;
 751 }
 752
 753 sub htmlfn ($) {
 754         return shift().".".($config{aggregateinternal} ? "_aggregated" : $config{htmlext});
 755 }
 756
 757 my $aggregatelock;
 758
 759 sub lockaggregate () {
 760         # Take an exclusive lock to prevent multiple concurrent aggregators.
 761         # Returns true if the lock was aquired.
 762         if (! -d $config{wikistatedir}) {
 763                 mkdir($config{wikistatedir});
 764         }
 765         open($aggregatelock, '>', "$config{wikistatedir}/aggregatelock") ||
 766                 error ("cannot open to $config{wikistatedir}/aggregatelock: $!");
 767         if (! flock($aggregatelock, 2 | 4)) { # LOCK_EX | LOCK_NB
 768                 close($aggregatelock) || error("failed closing aggregatelock: $!");
 769                 return 0;
 770         }
 771         return 1;
 772 }
 773
 774 sub unlockaggregate () {
 775         return close($aggregatelock) if $aggregatelock;
 776         return;
 777 }
 778
 779 1