git.oblomov.eu Git - ikiwiki/blob - IkiWiki/Plugin/aggregate.pm

   1 #!/usr/bin/perl
   2 # Blog aggregation plugin.
   3 package IkiWiki::Plugin::aggregate;
   4
   5 use warnings;
   6 use strict;
   7 use IkiWiki 2.00;
   8 use HTML::Entities;
   9 use HTML::Parser;
  10 use HTML::Tagset;
  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 => "checkconfig", id => "aggregate", call => \&checkconfig);
  20         hook(type => "filter", id => "aggregate", call => \&filter);
  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 } # }}}
  25
  26 sub getopt () { #{{{
  27         eval q{use Getopt::Long};
  28         error($@) if $@;
  29         Getopt::Long::Configure('pass_through');
  30         GetOptions("aggregate" => \$config{aggregate});
  31 } #}}}
  32
  33 sub checkconfig () { #{{{
  34         if ($config{aggregate} && ! ($config{post_commit} &&
  35                                      IkiWiki::commit_hook_enabled())) {
  36                 # don't wait for the lock
  37                 IkiWiki::lockwiki(0) || exit 1;
  38
  39                 loadstate();
  40                 IkiWiki::loadindex();
  41                 aggregate();
  42                 expire();
  43                 savestate();
  44                 clearstate();
  45
  46                 IkiWiki::unlockwiki();
  47         }
  48 } #}}}
  49
  50 sub filter (@) { #{{{
  51         my %params=@_;
  52         my $page=$params{page};
  53
  54         loadstate(); # if not already loaded
  55         # Mark all feeds originating on this page as removable;
  56         # preprocess will unmark those that still exist.
  57         remove_feeds($page);
  58
  59         return $params{content};
  60 } # }}}
  61
  62 sub preprocess (@) { #{{{
  63         my %params=@_;
  64
  65         foreach my $required (qw{name url}) {
  66                 if (! exists $params{$required}) {
  67                         return "[[aggregate ".sprintf(gettext("missing %s parameter"), $required)."]]";
  68                 }
  69         }
  70
  71         my $feed={};
  72         my $name=$params{name};
  73         if (exists $feeds{$name}) {
  74                 $feed=$feeds{$name};
  75         }
  76         else {
  77                 $feeds{$name}=$feed;
  78         }
  79         $feed->{name}=$name;
  80         $feed->{sourcepage}=$params{page};
  81         $feed->{url}=$params{url};
  82         my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".IkiWiki::titlepage($params{name});
  83         $dir=~s/^\/+//;
  84         ($dir)=$dir=~/$config{wiki_file_regexp}/;
  85         $feed->{dir}=$dir;
  86         $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
  87         $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
  88         $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
  89         $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
  90         delete $feed->{remove};
  91         delete $feed->{expired};
  92         $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
  93         $feed->{numposts}=0 unless defined $feed->{numposts};
  94         $feed->{newposts}=0 unless defined $feed->{newposts};
  95         $feed->{message}=gettext("new feed") unless defined $feed->{message};
  96         $feed->{error}=0 unless defined $feed->{error};
  97         $feed->{tags}=[];
  98         while (@_) {
  99                 my $key=shift;
 100                 my $value=shift;
 101                 if ($key eq 'tag') {
 102                         push @{$feed->{tags}}, $value;
 103                 }
 104         }
 105
 106         return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
 107                ($feed->{error} ? "<em>" : "").$feed->{message}.
 108                ($feed->{error} ? "</em>" : "").
 109                " (".$feed->{numposts}." ".gettext("posts").
 110                ($feed->{newposts} ? "; ".$feed->{newposts}.
 111                                     " ".gettext("new") : "").
 112                ")";
 113 } # }}}
 114
 115 sub delete (@) { #{{{
 116         my @files=@_;
 117
 118         # Remove feed data for removed pages.
 119         foreach my $file (@files) {
 120                 my $page=pagename($file);
 121                 remove_feeds($page);
 122         }
 123 } #}}}
 124
 125 my $state_loaded=0;
 126 sub loadstate () { #{{{
 127         return if $state_loaded;
 128         if (-e "$config{wikistatedir}/aggregate") {
 129                 open (IN, "$config{wikistatedir}/aggregate" ||
 130                         die "$config{wikistatedir}/aggregate: $!");
 131                 while (<IN>) {
 132                         $_=IkiWiki::possibly_foolish_untaint($_);
 133                         chomp;
 134                         my $data={};
 135                         foreach my $i (split(/ /, $_)) {
 136                                 my ($field, $val)=split(/=/, $i, 2);
 137                                 if ($field eq "name" || $field eq "feed" ||
 138                                     $field eq "guid" || $field eq "message") {
 139                                         $data->{$field}=decode_entities($val, " \t\n");
 140                                 }
 141                                 elsif ($field eq "tag") {
 142                                         push @{$data->{tags}}, $val;
 143                                 }
 144                                 else {
 145                                         $data->{$field}=$val;
 146                                 }
 147                         }
 148
 149                         if (exists $data->{name}) {
 150                                 $feeds{$data->{name}}=$data;
 151                         }
 152                         elsif (exists $data->{guid}) {
 153                                 $guids{$data->{guid}}=$data;
 154                         }
 155                 }
 156
 157                 close IN;
 158
 159                 $state_loaded=1;
 160         }
 161 } #}}}
 162
 163 sub savestate () { #{{{
 164         eval q{use HTML::Entities};
 165         error($@) if $@;
 166         my $newfile="$config{wikistatedir}/aggregate.new";
 167         # TODO: This cleanup function could use improvement. Any newly
 168         # aggregated files are left behind unrecorded, and should be deleted.
 169         my $cleanup = sub { unlink($newfile) };
 170         open (OUT, ">$newfile") || error("open $newfile: $!", $cleanup);
 171         foreach my $data (values %feeds, values %guids) {
 172                 if ($data->{remove}) {
 173                         if ($data->{name}) {
 174                                 foreach my $guid (values %guids) {
 175                                         if ($guid->{feed} eq $data->{name}) {
 176                                                 $guid->{remove}=1;
 177                                         }
 178                                 }
 179                         }
 180                         else {
 181                                 unlink pagefile($data->{page});
 182                         }
 183                         next;
 184                 }
 185                 elsif ($data->{expired} && exists $data->{page}) {
 186                         unlink pagefile($data->{page});
 187                         delete $data->{page};
 188                         delete $data->{md5};
 189                 }
 190
 191                 my @line;
 192                 foreach my $field (keys %$data) {
 193                         if ($field eq "name" || $field eq "feed" ||
 194                             $field eq "guid" || $field eq "message") {
 195                                 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
 196                         }
 197                         elsif ($field eq "tags") {
 198                                 push @line, "tag=$_" foreach @{$data->{tags}};
 199                         }
 200                         else {
 201                                 push @line, "$field=".$data->{$field};
 202                         }
 203                 }
 204                 print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
 205         }
 206         close OUT || error("save $newfile: $!", $cleanup);
 207         rename($newfile, "$config{wikistatedir}/aggregate") ||
 208                 error("rename $newfile: $!", $cleanup);
 209 } #}}}
 210
 211 sub clearstate () { #{{{
 212         %feeds=();
 213         %guids=();
 214         $state_loaded=0;
 215 } #}}}
 216
 217 sub expire () { #{{{
 218         foreach my $feed (values %feeds) {
 219                 next unless $feed->{expireage} || $feed->{expirecount};
 220                 my $count=0;
 221                 foreach my $item (sort { $IkiWiki::pagectime{$b->{page}} <=> $IkiWiki::pagectime{$a->{page}} }
 222                                   grep { exists $_->{page} && $_->{feed} eq $feed->{name} && $IkiWiki::pagectime{$_->{page}} }
 223                                   values %guids) {
 224                         if ($feed->{expireage}) {
 225                                 my $days_old = (time - $IkiWiki::pagectime{$item->{page}}) / 60 / 60 / 24;
 226                                 if ($days_old > $feed->{expireage}) {
 227                                         debug(sprintf(gettext("expiring %s (%s days old)"),
 228                                                 $item->{page}, $days_old));
 229                                         $item->{expired}=1;
 230                                 }
 231                         }
 232                         elsif ($feed->{expirecount} &&
 233                                $count >= $feed->{expirecount}) {
 234                                 debug(sprintf(gettext("expiring %s"), $item->{page}));
 235                                 $item->{expired}=1;
 236                         }
 237                         else {
 238                                 $count++;
 239                         }
 240                 }
 241         }
 242 } #}}}
 243
 244 sub aggregate () { #{{{
 245         eval q{use XML::Feed};
 246         error($@) if $@;
 247         eval q{use URI::Fetch};
 248         error($@) if $@;
 249         eval q{use HTML::Entities};
 250         error($@) if $@;
 251
 252         foreach my $feed (values %feeds) {
 253                 next unless $config{rebuild} ||
 254                         time - $feed->{lastupdate} >= $feed->{updateinterval};
 255                 $feed->{lastupdate}=time;
 256                 $feed->{newposts}=0;
 257                 $feed->{message}=sprintf(gettext("processed ok at %s"),
 258                         displaytime($feed->{lastupdate}));
 259                 $feed->{error}=0;
 260                 $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
 261
 262                 debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
 263
 264                 if (! length $feed->{feedurl}) {
 265                         my @urls=XML::Feed->find_feeds($feed->{url});
 266                         if (! @urls) {
 267                                 $feed->{message}=sprintf(gettext("could not find feed at %s"), $feed->{url});
 268                                 $feed->{error}=1;
 269                                 debug($feed->{message});
 270                                 next;
 271                         }
 272                         $feed->{feedurl}=pop @urls;
 273                 }
 274                 my $res=URI::Fetch->fetch($feed->{feedurl});
 275                 if (! $res) {
 276                         $feed->{message}=URI::Fetch->errstr;
 277                         $feed->{error}=1;
 278                         debug($feed->{message});
 279                         next;
 280                 }
 281                 if ($res->status == URI::Fetch::URI_GONE()) {
 282                         $feed->{message}=gettext("feed not found");
 283                         $feed->{error}=1;
 284                         debug($feed->{message});
 285                         next;
 286                 }
 287                 my $content=$res->content;
 288                 my $f=eval{XML::Feed->parse(\$content)};
 289                 if ($@) {
 290                         # One common cause of XML::Feed crashing is a feed
 291                         # that contains invalid UTF-8 sequences. Convert
 292                         # feed to ascii to try to work around.
 293                         $feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)"));
 294                         $content=Encode::decode_utf8($content);
 295                         $f=eval{XML::Feed->parse(\$content)};
 296                 }
 297                 if ($@) {
 298                         $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
 299                         $feed->{error}=1;
 300                         debug($feed->{message});
 301                         next;
 302                 }
 303                 if (! $f) {
 304                         $feed->{message}=XML::Feed->errstr;
 305                         $feed->{error}=1;
 306                         debug($feed->{message});
 307                         next;
 308                 }
 309
 310                 foreach my $entry ($f->entries) {
 311                         add_page(
 312                                 feed => $feed,
 313                                 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
 314                                 link => $entry->link,
 315                                 content => defined $entry->content->body ? $entry->content->body : "",
 316                                 guid => defined $entry->id ? $entry->id : time."_".$feed->name,
 317                                 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
 318                         );
 319                 }
 320         }
 321 } #}}}
 322
 323 sub add_page (@) { #{{{
 324         my %params=@_;
 325
 326         my $feed=$params{feed};
 327         my $guid={};
 328         my $mtime;
 329         if (exists $guids{$params{guid}}) {
 330                 # updating an existing post
 331                 $guid=$guids{$params{guid}};
 332                 return if $guid->{expired};
 333         }
 334         else {
 335                 # new post
 336                 $guid->{guid}=$params{guid};
 337                 $guids{$params{guid}}=$guid;
 338                 $mtime=$params{ctime};
 339                 $feed->{numposts}++;
 340                 $feed->{newposts}++;
 341
 342                 # assign it an unused page
 343                 my $page=IkiWiki::titlepage($params{title});
 344                 # escape slashes and periods in title so it doesn't specify
 345                 # directory name or trigger ".." disallowing code.
 346                 $page=~s!([/.])!"__".ord($1)."__"!eg;
 347                 $page=$feed->{dir}."/".$page;
 348                 ($page)=$page=~/$config{wiki_file_regexp}/;
 349                 if (! defined $page || ! length $page) {
 350                         $page=$feed->{dir}."/item";
 351                 }
 352                 my $c="";
 353                 while (exists $IkiWiki::pagecase{lc $page.$c} ||
 354                        -e pagefile($page.$c)) {
 355                         $c++
 356                 }
 357
 358                 # Make sure that the file name isn't too long.
 359                 # NB: This doesn't check for path length limits.
 360                 my $max=POSIX::pathconf($config{srcdir}, &POSIX::_PC_NAME_MAX);
 361                 if (defined $max && length(htmlfn($page)) >= $max) {
 362                         $c="";
 363                         $page=$feed->{dir}."/item";
 364                         while (exists $IkiWiki::pagecase{lc $page.$c} ||
 365                                -e pagefile($page.$c)) {
 366                                 $c++
 367                         }
 368                 }
 369
 370                 $guid->{page}=$page;
 371                 debug(sprintf(gettext("creating new page %s"), $page));
 372         }
 373         $guid->{feed}=$feed->{name};
 374
 375         # To write or not to write? Need to avoid writing unchanged pages
 376         # to avoid unneccessary rebuilding. The mtime from rss cannot be
 377         # trusted; let's use a digest.
 378         eval q{use Digest::MD5 'md5_hex'};
 379         error($@) if $@;
 380         require Encode;
 381         my $digest=md5_hex(Encode::encode_utf8($params{content}));
 382         return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
 383         $guid->{md5}=$digest;
 384
 385         # Create the page.
 386         my $template=template("aggregatepost.tmpl", blind_cache => 1);
 387         $template->param(title => $params{title})
 388                 if defined $params{title} && length($params{title});
 389         $template->param(content => htmlescape(htmlabs($params{content}, $feed->{feedurl})));
 390         $template->param(name => $feed->{name});
 391         $template->param(url => $feed->{url});
 392         $template->param(permalink => urlabs($params{link}, $feed->{feedurl}))
 393                 if defined $params{link};
 394         if (ref $feed->{tags}) {
 395                 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
 396         }
 397         writefile(htmlfn($guid->{page}), $config{srcdir},
 398                 $template->output);
 399
 400         # Set the mtime, this lets the build process get the right creation
 401         # time on record for the new page.
 402         utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
 403 } #}}}
 404
 405 sub htmlescape ($) { #{{{
 406         # escape accidental wikilinks and preprocessor stuff
 407         my $html=shift;
 408         $html=~s/(?<!\\)\[\[/\\\[\[/g;
 409         return $html;
 410 } #}}}
 411
 412 sub urlabs ($$) { #{{{
 413         my $url=shift;
 414         my $urlbase=shift;
 415
 416         URI->new_abs($url, $urlbase)->as_string;
 417 } #}}}
 418
 419 sub htmlabs ($$) { #{{{
 420         # Convert links in html from relative to absolute.
 421         # Note that this is a heuristic, which is not specified by the rss
 422         # spec and may not be right for all feeds. Also, see Debian
 423         # bug #381359.
 424         my $html=shift;
 425         my $urlbase=shift;
 426
 427         my $ret="";
 428         my $p = HTML::Parser->new(api_version => 3);
 429         $p->handler(default => sub { $ret.=join("", @_) }, "text");
 430         $p->handler(start => sub {
 431                 my ($tagname, $pos, $text) = @_;
 432                 if (ref $HTML::Tagset::linkElements{$tagname}) {
 433                         while (4 <= @$pos) {
 434                                 # use attribute sets from right to left
 435                                 # to avoid invalidating the offsets
 436                                 # when replacing the values
 437                                 my($k_offset, $k_len, $v_offset, $v_len) =
 438                                         splice(@$pos, -4);
 439                                 my $attrname = lc(substr($text, $k_offset, $k_len));
 440                                 next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
 441                                 next unless $v_offset; # 0 v_offset means no value
 442                                 my $v = substr($text, $v_offset, $v_len);
 443                                 $v =~ s/^([\'\"])(.*)\1$/$2/;
 444                                 my $new_v=urlabs($v, $urlbase);
 445                                 $new_v =~ s/\"/&quot;/g; # since we quote with ""
 446                                 substr($text, $v_offset, $v_len) = qq("$new_v");
 447                         }
 448                 }
 449                 $ret.=$text;
 450         }, "tagname, tokenpos, text");
 451         $p->parse($html);
 452         $p->eof;
 453
 454         return $ret;
 455 } #}}}
 456
 457 sub remove_feeds () { #{{{
 458         my $page=shift;
 459
 460         my %removed;
 461         foreach my $id (keys %feeds) {
 462                 if ($feeds{$id}->{sourcepage} eq $page) {
 463                         $feeds{$id}->{remove}=1;
 464                         $removed{$id}=1;
 465                 }
 466         }
 467 } #}}}
 468
 469 sub pagefile ($) { #{{{
 470         my $page=shift;
 471
 472         return "$config{srcdir}/".htmlfn($page);
 473 } #}}}
 474
 475 sub htmlfn ($) { #{{{
 476         return shift().".html";
 477 } #}}}
 478
 479 1