oops
[ikiwiki] / 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 use HTML::Entities;
9
10 my %feeds;
11 my %guids;
12
13 sub import { #{{{
14         IkiWiki::hook(type => "getopt", id => "aggregate", 
15                 call => \&getopt);
16         IkiWiki::hook(type => "checkconfig", id => "aggregate",
17                 call => \&checkconfig);
18         IkiWiki::hook(type => "filter", id => "aggregate", 
19                 call => \&filter);
20         IkiWiki::hook(type => "preprocess", id => "aggregate",
21                 call => \&preprocess);
22         IkiWiki::hook(type => "delete", id => "aggregate",
23                 call => \&delete);
24         IkiWiki::hook(type => "savestate", id => "aggregate",
25                 call => \&savestate);
26 } # }}}
27
28 sub getopt () { #{{{
29         eval q{use Getopt::Long};
30         Getopt::Long::Configure('pass_through');
31         GetOptions("aggregate" => \$IkiWiki::config{aggregate});
32 } #}}}
33
34 sub checkconfig () { #{{{
35         loadstate();
36         if ($IkiWiki::config{aggregate}) {
37                 IkiWiki::loadindex();
38                 aggregate();
39                 savestate();
40         }
41 } #}}}
42
43 sub filter (@) { #{{{
44         my %params=@_;
45         my $page=$params{page};
46
47         # Mark all feeds originating on this page as removable;
48         # preprocess will unmark those that still exist.
49         remove_feeds($page);
50
51         return $params{content};
52 } # }}}
53
54 sub preprocess (@) { #{{{
55         my %params=@_;
56
57         foreach my $required (qw{name url}) {
58                 if (! exists $params{$required}) {
59                         return "[[aggregate plugin missing $required parameter]]";
60                 }
61         }
62
63         my $feed={};
64         my $name=$params{name};
65         if (exists $feeds{$name}) {
66                 $feed=$feeds{$name};
67         }
68         else {
69                 $feeds{$name}=$feed;
70         }
71         $feed->{name}=$name;
72         $feed->{sourcepage}=$params{page};
73         $feed->{url}=$params{url};
74         my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".IkiWiki::titlepage($params{name});
75         $dir=~s/^\/+//;
76         ($dir)=$dir=~/$IkiWiki::config{wiki_file_regexp}/;
77         $feed->{dir}=$dir;
78         $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
79         $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
80         $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
81         $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
82         delete $feed->{remove};
83         $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
84         $feed->{numposts}=0 unless defined $feed->{numposts};
85         $feed->{newposts}=0 unless defined $feed->{newposts};
86         $feed->{message}="new feed" unless defined $feed->{message};
87         $feed->{tags}=[];
88         while (@_) {
89                 my $key=shift;
90                 my $value=shift;
91                 if ($key eq 'tag') {
92                         push @{$feed->{tags}}, $value;
93                 }
94         }
95
96         return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
97                "<i>".$feed->{message}."</i> (".$feed->{numposts}.
98                " stored posts; ".$feed->{newposts}." new)<br />";
99 } # }}}
100
101 sub delete (@) { #{{{
102         my @files=@_;
103
104         # Remove feed data for removed pages.
105         foreach my $file (@files) {
106                 my $page=IkiWiki::pagename($file);
107                 remove_feeds($page);
108         }
109 } #}}}
110
111 sub loadstate () { #{{{
112         if (-e "$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 $IkiWiki::config{rebuild} || 
191                         time - $feed->{lastupdate} >= $feed->{updateinterval};
192                 $feed->{lastupdate}=time;
193                 $feed->{newposts}=0;
194                 $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
195
196                 IkiWiki::debug("checking feed ".$feed->{name}." ...");
197
198                 if (! length $feed->{feedurl}) {
199                         my @urls=XML::Feed->find_feeds($feed->{url});
200                         if (! @urls) {
201                                 $feed->{message}="could not find feed at ".$feed->{feedurl};
202                                 IkiWiki::debug($feed->{message});
203                                 next;
204                         }
205                         $feed->{feedurl}=pop @urls;
206                 }
207                 my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))};
208                 if ($@) {
209                         $feed->{message}="feed crashed XML::Feed! $@";
210                         IkiWiki::debug($feed->{message});
211                         next;
212                 }
213                 if (! $f) {
214                         $feed->{message}=XML::Feed->errstr;
215                         IkiWiki::debug($feed->{message});
216                         next;
217                 }
218
219                 foreach my $entry ($f->entries) {
220                         add_page(
221                                 feed => $feed,
222                                 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
223                                 link => $entry->link,
224                                 content => $entry->content->body,
225                                 guid => defined $entry->id ? $entry->id : time."_".$feed->name,
226                                 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
227                         );
228                 }
229
230                 $feed->{message}="processed ok";
231         }
232
233         # TODO: expiry
234 } #}}}
235
236 sub add_page (@) { #{{{
237         my %params=@_;
238         
239         my $feed=$params{feed};
240         my $guid={};
241         my $mtime;
242         if (exists $guids{$params{guid}}) {
243                 # updating an existing post
244                 $guid=$guids{$params{guid}};
245         }
246         else {
247                 # new post
248                 $guid->{guid}=$params{guid};
249                 $guids{$params{guid}}=$guid;
250                 $mtime=$params{ctime};
251                 $feed->{numposts}++;
252                 $feed->{newposts}++;
253
254                 # assign it an unused page
255                 my $page=IkiWiki::titlepage($params{title});
256                 # escape slashes and periods in title so it doesn't specify
257                 # directory name or trigger ".." disallowing code.
258                 $page=~s!([/.])!"__".ord($1)."__"!eg;
259                 $page=$feed->{dir}."/".$page;
260                 $page=lc($page);
261                 ($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
262                 if (! defined $page || ! length $page) {
263                         $page=$feed->{dir}."/item";
264                 }
265                 my $c="";
266                 while (exists $IkiWiki::pagesources{$page.$c} ||
267                        -e pagefile($page.$c)) {
268                         $c++
269                 }
270                 $guid->{page}=$page;
271                 IkiWiki::debug("creating new page $page");
272         }
273         $guid->{feed}=$feed->{name};
274         
275         # To write or not to write? Need to avoid writing unchanged pages
276         # to avoid unneccessary rebuilding. The mtime from rss cannot be
277         # trusted; let's use a digest.
278         eval q{use Digest::MD5 'md5_hex'};
279         require Encode;
280         my $digest=md5_hex(Encode::encode_utf8($params{content}));
281         return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $IkiWiki::config{rebuild};
282         $guid->{md5}=$digest;
283
284         # Create the page.
285         my $template=IkiWiki::template("aggregatepost.tmpl", blind_cache => 1);
286         my $content=$params{content};
287         $params{content}=~s/(?<!\\)\[\[/\\\[\[/g; # escape accidental wikilinks
288                                                   # and preprocessor stuff
289         $template->param(title => $params{title})
290                 if defined $params{title} && length($params{title});
291         $template->param(content => $params{content});
292         $template->param(url => $feed->{url});
293         $template->param(name => $feed->{name});
294         $template->param(link => $params{link}) if defined $params{link};
295         if (ref $feed->{tags}) {
296                 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
297         }
298         IkiWiki::writefile($guid->{page}.".html", $IkiWiki::config{srcdir},
299                 $template->output);
300
301         # Set the mtime, this lets the build process get the right creation
302         # time on record for the new page.
303         utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
304 } #}}}
305
306 sub remove_feeds () { #{{{
307         my $page=shift;
308
309         my %removed;
310         foreach my $id (keys %feeds) {
311                 if ($feeds{$id}->{sourcepage} eq $page) {
312                         $feeds{$id}->{remove}=1;
313                         $removed{$id}=1;
314                 }
315         }
316 } #}}}
317
318 sub pagefile ($) { #{{{
319         my $page=shift;
320
321         return "$IkiWiki::config{srcdir}/$page.html";
322 } #}}}
323
324 1