summaryrefslogtreecommitdiff
path: root/IkiWiki/Plugin/aggregate.pm
blob: 99b26a546e91adb9ed35a7fea7ecb860884d8868 (plain)
  1. #!/usr/bin/perl
  2. # Blog aggregation plugin.
  3. package IkiWiki::Plugin::aggregate;
  4. use warnings;
  5. use strict;
  6. use IkiWiki;
  7. my %feeds;
  8. my %guids;
  9. sub import { #{{{
  10. IkiWiki::hook(type => "getopt", id => "aggregate",
  11. call => \&getopt);
  12. IkiWiki::hook(type => "checkconfig", id => "aggregate",
  13. call => \&checkconfig);
  14. IkiWiki::hook(type => "filter", id => "aggregate",
  15. call => \&filter);
  16. IkiWiki::hook(type => "preprocess", id => "aggregate",
  17. call => \&preprocess);
  18. IkiWiki::hook(type => "delete", id => "aggregate",
  19. call => \&delete);
  20. IkiWiki::hook(type => "savestate", id => "aggregate",
  21. call => \&savestate);
  22. } # }}}
  23. sub getopt () { #{{{
  24. eval q{use Getopt::Long};
  25. Getopt::Long::Configure('pass_through');
  26. GetOptions("aggregate" => \$IkiWiki::config{aggregate});
  27. } #}}}
  28. sub checkconfig () { #{{{
  29. loadstate();
  30. if ($IkiWiki::config{aggregate}) {
  31. IkiWiki::loadindex();
  32. aggregate();
  33. savestate();
  34. }
  35. } #}}}
  36. sub filter (@) { #{{{
  37. my %params=@_;
  38. my $page=$params{page};
  39. # Mark all feeds originating on this page as removable;
  40. # preprocess will unmark those that still exist.
  41. remove_feeds($page);
  42. return $params{content};
  43. } # }}}
  44. sub preprocess (@) { #{{{
  45. my %params=@_;
  46. foreach my $required (qw{name url}) {
  47. if (! exists $params{$required}) {
  48. return "[[aggregate plugin missing $required parameter]]";
  49. }
  50. }
  51. my $feed={};
  52. my $name=$params{name};
  53. if (exists $feeds{$name}) {
  54. $feed=$feeds{$name};
  55. }
  56. else {
  57. $feeds{$name}=$feed;
  58. }
  59. $feed->{name}=$name;
  60. $feed->{sourcepage}=$params{page};
  61. $feed->{url}=$params{url};
  62. my $dir=exists $params{dir} ? $params{dir} : IkiWiki::titlepage($params{name});
  63. $dir=~s/^\/+//;
  64. ($dir)=$dir=~/$IkiWiki::config{wiki_file_regexp}/;
  65. $feed->{dir}=$dir;
  66. $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : $params{url};
  67. $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
  68. $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
  69. $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
  70. delete $feed->{remove};
  71. $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
  72. $feed->{numposts}=0 unless defined $feed->{numposts};
  73. $feed->{newposts}=0 unless defined $feed->{newposts};
  74. $feed->{message}="new feed" unless defined $feed->{message};
  75. while (@_) {
  76. my $key=shift;
  77. my $value=shift;
  78. if ($key eq 'tag') {
  79. push @{$feed->{tags}}, $value;
  80. }
  81. }
  82. return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
  83. "<i>".$feed->{message}."</i> (".$feed->{numposts}.
  84. " stored posts; ".$feed->{newposts}." new)";
  85. } # }}}
  86. sub delete (@) { #{{{
  87. my @files=@_;
  88. # Remove feed data for removed pages.
  89. foreach my $file (@files) {
  90. my $page=IkiWiki::pagename($file);
  91. remove_feeds($page);
  92. }
  93. } #}}}
  94. sub loadstate () { #{{{
  95. if (-e "$IkiWiki::config{wikistatedir}/aggregate") {
  96. open (IN, "$IkiWiki::config{wikistatedir}/aggregate" ||
  97. die "$IkiWiki::config{wikistatedir}/aggregate: $!");
  98. while (<IN>) {
  99. $_=IkiWiki::possibly_foolish_untaint($_);
  100. chomp;
  101. my $data={};
  102. foreach my $i (split(/ /, $_)) {
  103. my ($field, $val)=split(/=/, $i, 2);
  104. if ($field eq "name" || $field eq "message") {
  105. $data->{$field}=IkiWiki::pagetitle($val);
  106. }
  107. elsif ($field eq "tag") {
  108. push @{$data->{tags}}, $val;
  109. }
  110. else {
  111. $data->{$field}=$val;
  112. }
  113. }
  114. if (exists $data->{name}) {
  115. $feeds{$data->{name}}=$data;
  116. }
  117. elsif (exists $data->{guid}) {
  118. $guids{$data->{guid}}=$data;
  119. }
  120. }
  121. close IN;
  122. }
  123. } #}}}
  124. sub savestate () { #{{{
  125. open (OUT, ">$IkiWiki::config{wikistatedir}/aggregate" ||
  126. die "$IkiWiki::config{wikistatedir}/aggregate: $!");
  127. foreach my $data (values %feeds, values %guids) {
  128. if ($data->{remove}) {
  129. if ($data->{name}) {
  130. foreach my $guid (values %guids) {
  131. if ($guid->{feed} eq $data->{name}) {
  132. $guid->{remove}=1;
  133. }
  134. }
  135. }
  136. else {
  137. unlink pagefile($data->{page});
  138. }
  139. next;
  140. }
  141. my @line;
  142. foreach my $field (keys %$data) {
  143. if ($field eq "name" || $field eq "message") {
  144. push @line, "$field=".IkiWiki::titlepage($data->{$field});
  145. }
  146. elsif ($field eq "tags") {
  147. push @line, "tag=$_" foreach @{$data->{tags}};
  148. }
  149. else {
  150. push @line, "$field=".$data->{$field};
  151. }
  152. }
  153. print OUT join(" ", @line)."\n";
  154. }
  155. close OUT;
  156. } #}}}
  157. sub aggregate () { #{{{
  158. eval q{use XML::Feed};
  159. die $@ if $@;
  160. eval q{use HTML::Entities};
  161. die $@ if $@;
  162. FEED: foreach my $feed (values %feeds) {
  163. next unless time - $feed->{lastupdate} >= $feed->{updateinterval};
  164. $feed->{lastupdate}=time;
  165. $feed->{newposts}=0;
  166. $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
  167. IkiWiki::debug("checking feed ".$feed->{name}." ...");
  168. my @urls=XML::Feed->find_feeds($feed->{feedurl});
  169. if (! @urls) {
  170. $feed->{message}="could not find feed at ".$feed->{feedurl};
  171. IkiWiki::debug($feed->{message});
  172. }
  173. foreach my $url (@urls) {
  174. my $f=XML::Feed->parse(URI->new($url));
  175. if (! $f) {
  176. $feed->{message}=XML::Feed->errstr;
  177. IkiWiki::debug($feed->{message});
  178. next FEED;
  179. }
  180. foreach my $entry ($f->entries) {
  181. add_page(
  182. feed => $feed,
  183. title => defined $entry->title ? decode_entities($entry->title) : "untitled",
  184. link => $entry->link,
  185. content => $entry->content->body,
  186. guid => defined $entry->id ? $entry->id : time."_".$feed->name,
  187. ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
  188. );
  189. }
  190. }
  191. $feed->{message}="processed ok";
  192. }
  193. # TODO: expiry
  194. } #}}}
  195. sub add_page (@) { #{{{
  196. my %params=@_;
  197. my $feed=$params{feed};
  198. my $guid={};
  199. my $mtime;
  200. if (exists $guids{$params{guid}}) {
  201. # updating an existing post
  202. $guid=$guids{$params{guid}};
  203. }
  204. else {
  205. # new post
  206. $guid->{guid}=$params{guid};
  207. $guids{$params{guid}}=$guid;
  208. $mtime=$params{ctime};
  209. $feed->{numposts}++;
  210. $feed->{newposts}++;
  211. # assign it an unused page
  212. my $page=$feed->{dir}."/".IkiWiki::titlepage($params{title});
  213. ($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
  214. if (! defined $page || ! length $page) {
  215. $page=$feed->{dir}."/item";
  216. }
  217. $page=~s/\.\.//g; # avoid ".." directory tricks
  218. my $c="";
  219. while (exists $IkiWiki::pagesources{$page.$c} ||
  220. -e pagefile($page.$c)) {
  221. $c++
  222. }
  223. $guid->{page}=$page;
  224. IkiWiki::debug("creating new page $page");
  225. }
  226. $guid->{feed}=$feed->{name};
  227. # To write or not to write? Need to avoid writing unchanged pages
  228. # to avoid unneccessary rebuilding. The mtime from rss cannot be
  229. # trusted; let's use a digest.
  230. eval q{use Digest::MD5 'md5_hex'};
  231. require Encode;
  232. my $digest=md5_hex(Encode::encode_utf8($params{content}));
  233. return unless ! exists $guid->{md5} || $guid->{md5} ne $digest;
  234. $guid->{md5}=$digest;
  235. # Create the page.
  236. my $template=IkiWiki::template("aggregatepost.tmpl", blind_cache => 1);
  237. my $content=$params{content};
  238. $params{content}=~s/(?<!\\)\[\[/\\\[\[/g; # escape accidental wikilinks
  239. # and preprocessor stuff
  240. $template->param(content => $params{content});
  241. $template->param(url => $feed->{url});
  242. $template->param(name => $feed->{name});
  243. $template->param(link => $params{link}) if defined $params{link};
  244. if (ref $feed->{tags}) {
  245. $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
  246. }
  247. IkiWiki::writefile($guid->{page}.".html", $IkiWiki::config{srcdir},
  248. $template->output);
  249. # Set the mtime, this lets the build process get the right creation
  250. # time on record for the new page.
  251. utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
  252. } #}}}
  253. sub remove_feeds () { #{{{
  254. my $page=shift;
  255. my %removed;
  256. foreach my $id (keys %feeds) {
  257. if ($feeds{$id}->{sourcepage} eq $page) {
  258. $feeds{$id}->{remove}=1;
  259. $removed{$id}=1;
  260. }
  261. }
  262. } #}}}
  263. sub pagefile ($) { #{{{
  264. my $page=shift;
  265. return "$IkiWiki::config{srcdir}/$page.html";
  266. } #}}}
  267. 1