summaryrefslogtreecommitdiff
path: root/IkiWiki/Plugin/aggregate.pm
blob: 5c8151b58830709ae4172a194eb2cd6cc24894fc (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 dir}) {
  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. ($feed->{dir})=$params{dir}=~/$IkiWiki::config{wiki_file_regexp}/;
  63. $feed->{dir}=~s/^\/+//;
  64. $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : $params{url};
  65. $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} : 15;
  66. $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
  67. $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
  68. delete $feed->{remove};
  69. $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
  70. $feed->{numposts}=0 unless defined $feed->{numposts};
  71. $feed->{newposts}=0 unless defined $feed->{newposts};
  72. $feed->{message}="new feed" unless defined $feed->{message};
  73. while (@_) {
  74. my $key=shift;
  75. my $value=shift;
  76. if ($key eq 'tag') {
  77. push @{$feed->{tags}}, $value;
  78. }
  79. }
  80. return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
  81. "<i>".$feed->{message}."</i> (".$feed->{numposts}.
  82. " stored posts; ".$feed->{newposts}." new)";
  83. } # }}}
  84. sub delete (@) { #{{{
  85. my @files=@_;
  86. # Remove feed data for removed pages.
  87. foreach my $file (@files) {
  88. my $page=IkiWiki::pagename($file);
  89. remove_feeds($page);
  90. }
  91. } #}}}
  92. sub loadstate () { #{{{
  93. if (-e "$IkiWiki::config{wikistatedir}/aggregate") {
  94. open (IN, "$IkiWiki::config{wikistatedir}/aggregate" ||
  95. die "$IkiWiki::config{wikistatedir}/aggregate: $!");
  96. while (<IN>) {
  97. $_=IkiWiki::possibly_foolish_untaint($_);
  98. chomp;
  99. my $data={};
  100. foreach my $i (split(/ /, $_)) {
  101. my ($field, $val)=split(/=/, $i, 2);
  102. if ($field eq "name" || $field eq "message") {
  103. $data->{$field}=IkiWiki::pagetitle($val);
  104. }
  105. elsif ($field eq "tag") {
  106. push @{$data->{tags}}, $val;
  107. }
  108. else {
  109. $data->{$field}=$val;
  110. }
  111. }
  112. if (exists $data->{name}) {
  113. $feeds{$data->{name}}=$data;
  114. }
  115. elsif (exists $data->{guid}) {
  116. $guids{$data->{guid}}=$data;
  117. }
  118. }
  119. close IN;
  120. }
  121. } #}}}
  122. sub savestate () { #{{{
  123. open (OUT, ">$IkiWiki::config{wikistatedir}/aggregate" ||
  124. die "$IkiWiki::config{wikistatedir}/aggregate: $!");
  125. foreach my $data (values %feeds, values %guids) {
  126. if ($data->{remove}) {
  127. if ($data->{name}) {
  128. foreach my $guid (values %guids) {
  129. if ($guid->{feed} eq $data->{name}) {
  130. $guid->{remove}=1;
  131. }
  132. }
  133. }
  134. else {
  135. unlink pagefile($data->{page});
  136. }
  137. next;
  138. }
  139. my @line;
  140. foreach my $field (keys %$data) {
  141. if ($field eq "name" || $field eq "message") {
  142. push @line, "$field=".IkiWiki::titlepage($data->{$field});
  143. }
  144. elsif ($field eq "tags") {
  145. push @line, "tag=$_" foreach @{$data->{tags}};
  146. }
  147. else {
  148. push @line, "$field=".$data->{$field};
  149. }
  150. }
  151. print OUT join(" ", @line)."\n";
  152. }
  153. close OUT;
  154. } #}}}
  155. sub aggregate () { #{{{
  156. eval q{use XML::Feed};
  157. die $@ if $@;
  158. eval q{use HTML::Entities};
  159. die $@ if $@;
  160. FEED: foreach my $feed (values %feeds) {
  161. # TODO: check updateinterval
  162. $feed->{lastupdate}=time;
  163. $feed->{newposts}=0;
  164. $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
  165. IkiWiki::debug("checking feed ".$feed->{name}." ...");
  166. my @urls=XML::Feed->find_feeds($feed->{feedurl});
  167. if (! @urls) {
  168. $feed->{message}="could not find feed at ".$feed->{feedurl};
  169. IkiWiki::debug($feed->{message});
  170. }
  171. foreach my $url (@urls) {
  172. my $f=XML::Feed->parse(URI->new($url));
  173. if (! $f) {
  174. $feed->{message}=XML::Feed->errstr;
  175. IkiWiki::debug($feed->{message});
  176. next FEED;
  177. }
  178. foreach my $entry ($f->entries) {
  179. add_page(
  180. feed => $feed,
  181. title => decode_entities($entry->title),
  182. link => $entry->link,
  183. content => $entry->content->body,
  184. guid => defined $entry->id ? $entry->id : time."_".$feed->name,
  185. ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
  186. );
  187. }
  188. }
  189. $feed->{message}="processed ok";
  190. }
  191. # TODO: expiry
  192. } #}}}
  193. sub add_page (@) { #{{{
  194. my %params=@_;
  195. my $feed=$params{feed};
  196. my $guid={};
  197. my $mtime;
  198. if (exists $guids{$params{guid}}) {
  199. # updating an existing post
  200. $guid=$guids{$params{guid}};
  201. }
  202. else {
  203. # new post
  204. $guid->{guid}=$params{guid};
  205. $guids{$params{guid}}=$guid;
  206. $mtime=$params{ctime};
  207. $feed->{numposts}++;
  208. $feed->{newposts}++;
  209. # assign it an unused page
  210. my $page=$feed->{dir}."/".IkiWiki::titlepage($params{title});
  211. ($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
  212. if (! defined $page || ! length $page) {
  213. $page=$feed->{dir}."/item";
  214. }
  215. $page=~s/\.\.//g; # avoid ".." directory tricks
  216. my $c="";
  217. while (exists $IkiWiki::pagesources{$page.$c} ||
  218. -e pagefile($page.$c)) {
  219. $c++
  220. }
  221. $guid->{page}=$page;
  222. IkiWiki::debug("creating new page $page");
  223. }
  224. $guid->{feed}=$feed->{name};
  225. # To write or not to write? Need to avoid writing unchanged pages
  226. # to avoid unneccessary rebuilding. The mtime from rss cannot be
  227. # trusted; let's use a digest.
  228. eval q{use Digest::MD5 'md5_hex'};
  229. my $digest=md5_hex($params{content});
  230. return unless ! exists $guid->{md5} || $guid->{md5} ne $digest;
  231. $guid->{md5}=$digest;
  232. # Create the page.
  233. my $template=IkiWiki::template("aggregatepost.tmpl", blind_cache => 1);
  234. my $content=$params{content};
  235. $params{content}=~s/(?<!\\)\[\[/\\\[\[/g; # escape accidental wikilinks
  236. # and preprocessor stuff
  237. $template->param(content => $params{content});
  238. $template->param(url => $feed->{url});
  239. $template->param(name => $feed->{name});
  240. $template->param(link => $params{link}) if defined $params{link};
  241. if (ref $feed->{tags}) {
  242. $template->param(tags => map { tag => $_ }, @{$feed->{tags}});
  243. }
  244. IkiWiki::writefile($guid->{page}.".html", $IkiWiki::config{srcdir},
  245. $template->output);
  246. # Set the mtime, this lets the build process get the right creation
  247. # time on record for the new page.
  248. utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
  249. } #}}}
  250. sub remove_feeds () { #{{{
  251. my $page=shift;
  252. my %removed;
  253. foreach my $id (keys %feeds) {
  254. if ($feeds{$id}->{sourcepage} eq $page) {
  255. $feeds{$id}->{remove}=1;
  256. $removed{$id}=1;
  257. }
  258. }
  259. } #}}}
  260. sub pagefile ($) { #{{{
  261. my $page=shift;
  262. return "$IkiWiki::config{srcdir}/$page.html";
  263. } #}}}
  264. 1