]> sipb.mit.edu Git - ikiwiki.git/blob - IkiWiki/Plugin/aggregate.pm
41ca9cce0f7ff2f4029cd123102d1b04276f96f5
[ikiwiki.git] / 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;
8
9 my %feeds;
10 my %guids;
11
12 sub import { #{{{
13         IkiWiki::hook(type => "getopt", id => "aggregate", 
14                 call => \&getopt);
15         IkiWiki::hook(type => "checkconfig", id => "aggregate",
16                 call => \&checkconfig);
17         IkiWiki::hook(type => "filter", id => "aggregate", 
18                 call => \&filter);
19         IkiWiki::hook(type => "preprocess", id => "aggregate",
20                 call => \&preprocess);
21         IkiWiki::hook(type => "delete", id => "aggregate",
22                 call => \&delete);
23         IkiWiki::hook(type => "savestate", id => "aggregate",
24                 call => \&savestate);
25 } # }}}
26
27 sub getopt () { #{{{
28         eval q{use Getopt::Long};
29         Getopt::Long::Configure('pass_through');
30         GetOptions("aggregate" => \$IkiWiki::config{aggregate});
31 } #}}}
32
33 sub checkconfig () { #{{{
34         loadstate();
35         if ($IkiWiki::config{aggregate}) {
36                 IkiWiki::loadindex();
37                 aggregate();
38                 savestate();
39         }
40 } #}}}
41
42 sub filter (@) { #{{{
43         my %params=@_;
44         my $page=$params{page};
45
46         # Mark all feeds originating on this page as removable;
47         # preprocess will unmark those that still exist.
48         remove_feeds($page);
49
50         return $params{content};
51 } # }}}
52
53 sub preprocess (@) { #{{{
54         my %params=@_;
55
56         foreach my $required (qw{name url}) {
57                 if (! exists $params{$required}) {
58                         return "[[aggregate plugin missing $required parameter]]";
59                 }
60         }
61
62         my $feed={};
63         my $name=$params{name};
64         if (exists $feeds{$name}) {
65                 $feed=$feeds{$name};
66         }
67         else {
68                 $feeds{$name}=$feed;
69         }
70         $feed->{name}=$name;
71         $feed->{sourcepage}=$params{page};
72         $feed->{url}=$params{url};
73         my $dir=exists $params{dir} ? $params{dir} : IkiWiki::titlepage($params{name});
74         $dir=~s/^\/+//;
75         ($dir)=$dir=~/$IkiWiki::config{wiki_file_regexp}/;
76         $feed->{dir}=$dir;
77         $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
78         $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
79         $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
80         $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
81         delete $feed->{remove};
82         $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
83         $feed->{numposts}=0 unless defined $feed->{numposts};
84         $feed->{newposts}=0 unless defined $feed->{newposts};
85         $feed->{message}="new feed" unless defined $feed->{message};
86         while (@_) {
87                 my $key=shift;
88                 my $value=shift;
89                 if ($key eq 'tag') {
90                         push @{$feed->{tags}}, $value;
91                 }
92         }
93
94         return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
95                "<i>".$feed->{message}."</i> (".$feed->{numposts}.
96                " stored posts; ".$feed->{newposts}." new)<br />";
97 } # }}}
98
99 sub delete (@) { #{{{
100         my @files=@_;
101
102         # Remove feed data for removed pages.
103         foreach my $file (@files) {
104                 my $page=IkiWiki::pagename($file);
105                 remove_feeds($page);
106         }
107 } #}}}
108
109 sub loadstate () { #{{{
110         eval q{use HTML::Entities};
111         die $@ if $@;
112         if (-e "$IkiWiki::config{wikistatedir}/aggregate") {
113                 open (IN, "$IkiWiki::config{wikistatedir}/aggregate" ||
114                         die "$IkiWiki::config{wikistatedir}/aggregate: $!");
115                 while (<IN>) {
116                         $_=IkiWiki::possibly_foolish_untaint($_);
117                         chomp;
118                         my $data={};
119                         foreach my $i (split(/ /, $_)) {
120                                 my ($field, $val)=split(/=/, $i, 2);
121                                 if ($field eq "name" || $field eq "feed" ||
122                                     $field eq "guid" || $field eq "message") {
123                                         $data->{$field}=decode_entities($val, " \t\n");
124                                 }
125                                 elsif ($field eq "tag") {
126                                         push @{$data->{tags}}, $val;
127                                 }
128                                 else {
129                                         $data->{$field}=$val;
130                                 }
131                         }
132                         
133                         if (exists $data->{name}) {
134                                 $feeds{$data->{name}}=$data;
135                         }
136                         elsif (exists $data->{guid}) {
137                                 $guids{$data->{guid}}=$data;
138                         }
139                 }
140
141                 close IN;
142         }
143 } #}}}
144
145 sub savestate () { #{{{
146         eval q{use HTML::Entities};
147         die $@ if $@;
148         open (OUT, ">$IkiWiki::config{wikistatedir}/aggregate" ||
149                 die "$IkiWiki::config{wikistatedir}/aggregate: $!");
150         foreach my $data (values %feeds, values %guids) {
151                 if ($data->{remove}) {
152                         if ($data->{name}) {
153                                 foreach my $guid (values %guids) {
154                                         if ($guid->{feed} eq $data->{name}) {
155                                                 $guid->{remove}=1;
156                                         }
157                                 }
158                         }
159                         else {
160                                 unlink pagefile($data->{page});
161                         }
162                         next;
163                 }
164
165                 my @line;
166                 foreach my $field (keys %$data) {
167                         if ($field eq "name" || $field eq "feed" ||
168                             $field eq "guid" || $field eq "message") {
169                                 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
170                         }
171                         elsif ($field eq "tags") {
172                                 push @line, "tag=$_" foreach @{$data->{tags}};
173                         }
174                         else {
175                                 push @line, "$field=".$data->{$field};
176                         }
177                 }
178                 print OUT join(" ", @line)."\n";
179         }
180         close OUT;
181 } #}}}
182
183 sub aggregate () { #{{{
184         eval q{use XML::Feed};
185         die $@ if $@;
186         eval q{use HTML::Entities};
187         die $@ if $@;
188
189         foreach my $feed (values %feeds) {
190                 next unless time - $feed->{lastupdate} >= $feed->{updateinterval};
191                 $feed->{lastupdate}=time;
192                 $feed->{newposts}=0;
193                 $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
194
195                 IkiWiki::debug("checking feed ".$feed->{name}." ...");
196
197                 if (! length $feed->{feedurl}) {
198                         my @urls=XML::Feed->find_feeds($feed->{url});
199                         if (! @urls) {
200                                 $feed->{message}="could not find feed at ".$feed->{feedurl};
201                                 IkiWiki::debug($feed->{message});
202                                 next;
203                         }
204                         $feed->{feedurl}=pop @urls;
205                 }
206                 my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))};
207                 if ($@) {
208                         $feed->{message}="feed crashed XML::Feed! $@";
209                         IkiWiki::debug($feed->{message});
210                         next;
211                 }
212                 if (! $f) {
213                         $feed->{message}=XML::Feed->errstr;
214                         IkiWiki::debug($feed->{message});
215                         next;
216                 }
217
218                 foreach my $entry ($f->entries) {
219                         add_page(
220                                 feed => $feed,
221                                 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
222                                 link => $entry->link,
223                                 content => $entry->content->body,
224                                 guid => defined $entry->id ? $entry->id : time."_".$feed->name,
225                                 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
226                         );
227                 }
228
229                 $feed->{message}="processed ok";
230         }
231
232         # TODO: expiry
233 } #}}}
234
235 sub add_page (@) { #{{{
236         my %params=@_;
237         
238         my $feed=$params{feed};
239         my $guid={};
240         my $mtime;
241         if (exists $guids{$params{guid}}) {
242                 # updating an existing post
243                 $guid=$guids{$params{guid}};
244         }
245         else {
246                 # new post
247                 $guid->{guid}=$params{guid};
248                 $guids{$params{guid}}=$guid;
249                 $mtime=$params{ctime};
250                 $feed->{numposts}++;
251                 $feed->{newposts}++;
252
253                 # assign it an unused page
254                 my $page=$feed->{dir}."/".IkiWiki::titlepage($params{title});
255                 ($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
256                 if (! defined $page || ! length $page) {
257                         $page=$feed->{dir}."/item";
258                 }
259                 $page=~s/\.\.//g; # avoid ".." directory tricks
260                 my $c="";
261                 while (exists $IkiWiki::pagesources{$page.$c} ||
262                        -e pagefile($page.$c)) {
263                         $c++
264                 }
265                 $guid->{page}=$page;
266                 IkiWiki::debug("creating new page $page");
267         }
268         $guid->{feed}=$feed->{name};
269         
270         # To write or not to write? Need to avoid writing unchanged pages
271         # to avoid unneccessary rebuilding. The mtime from rss cannot be
272         # trusted; let's use a digest.
273         eval q{use Digest::MD5 'md5_hex'};
274         require Encode;
275         my $digest=md5_hex(Encode::encode_utf8($params{content}));
276         return unless ! exists $guid->{md5} || $guid->{md5} ne $digest;
277         $guid->{md5}=$digest;
278
279         # Create the page.
280         my $template=IkiWiki::template("aggregatepost.tmpl", blind_cache => 1);
281         my $content=$params{content};
282         $params{content}=~s/(?<!\\)\[\[/\\\[\[/g; # escape accidental wikilinks
283                                                   # and preprocessor stuff
284         $template->param(content => $params{content});
285         $template->param(url => $feed->{url});
286         $template->param(name => $feed->{name});
287         $template->param(link => $params{link}) if defined $params{link};
288         if (ref $feed->{tags}) {
289                 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
290         }
291         IkiWiki::writefile($guid->{page}.".html", $IkiWiki::config{srcdir},
292                 $template->output);
293
294         # Set the mtime, this lets the build process get the right creation
295         # time on record for the new page.
296         utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
297 } #}}}
298
299 sub remove_feeds () { #{{{
300         my $page=shift;
301
302         my %removed;
303         foreach my $id (keys %feeds) {
304                 if ($feeds{$id}->{sourcepage} eq $page) {
305                         $feeds{$id}->{remove}=1;
306                         $removed{$id}=1;
307                 }
308         }
309 } #}}}
310
311 sub pagefile ($) { #{{{
312         my $page=shift;
313
314         return "$IkiWiki::config{srcdir}/$page.html";
315 } #}}}
316
317 1