remove the "0 new", it's not very useful
[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 use HTML::Parser;
10 use HTML::Tagset;
11 use URI;
12
13 my %feeds;
14 my %guids;
15
16 sub import { #{{{
17         IkiWiki::hook(type => "getopt", id => "aggregate", 
18                 call => \&getopt);
19         IkiWiki::hook(type => "checkconfig", id => "aggregate",
20                 call => \&checkconfig);
21         IkiWiki::hook(type => "filter", id => "aggregate", 
22                 call => \&filter);
23         IkiWiki::hook(type => "preprocess", id => "aggregate",
24                 call => \&preprocess);
25         IkiWiki::hook(type => "delete", id => "aggregate",
26                 call => \&delete);
27         IkiWiki::hook(type => "savestate", id => "aggregate",
28                 call => \&savestate);
29 } # }}}
30
31 sub getopt () { #{{{
32         eval q{use Getopt::Long};
33         Getopt::Long::Configure('pass_through');
34         GetOptions("aggregate" => \$IkiWiki::config{aggregate});
35 } #}}}
36
37 sub checkconfig () { #{{{
38         IkiWiki::lockwiki();
39         loadstate();
40         if ($IkiWiki::config{aggregate}) {
41                 IkiWiki::loadindex();
42                 aggregate();
43                 savestate();
44         }
45         IkiWiki::unlockwiki();
46 } #}}}
47
48 sub filter (@) { #{{{
49         my %params=@_;
50         my $page=$params{page};
51
52         # Mark all feeds originating on this page as removable;
53         # preprocess will unmark those that still exist.
54         remove_feeds($page);
55
56         return $params{content};
57 } # }}}
58
59 sub preprocess (@) { #{{{
60         my %params=@_;
61
62         foreach my $required (qw{name url}) {
63                 if (! exists $params{$required}) {
64                         return "[[aggregate plugin missing $required parameter]]";
65                 }
66         }
67
68         my $feed={};
69         my $name=$params{name};
70         if (exists $feeds{$name}) {
71                 $feed=$feeds{$name};
72         }
73         else {
74                 $feeds{$name}=$feed;
75         }
76         $feed->{name}=$name;
77         $feed->{sourcepage}=$params{page};
78         $feed->{url}=$params{url};
79         my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".IkiWiki::titlepage($params{name});
80         $dir=~s/^\/+//;
81         ($dir)=$dir=~/$IkiWiki::config{wiki_file_regexp}/;
82         $feed->{dir}=$dir;
83         $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
84         $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
85         $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
86         $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
87         delete $feed->{remove};
88         $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
89         $feed->{numposts}=0 unless defined $feed->{numposts};
90         $feed->{newposts}=0 unless defined $feed->{newposts};
91         $feed->{message}="new feed" unless defined $feed->{message};
92         $feed->{tags}=[];
93         while (@_) {
94                 my $key=shift;
95                 my $value=shift;
96                 if ($key eq 'tag') {
97                         push @{$feed->{tags}}, $value;
98                 }
99         }
100
101         return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
102                "<i>".$feed->{message}."</i> (".$feed->{numposts}.
103                " stored posts".
104                ($feed->{newposts} ? "; ".$feed->{newposts}." new" : "").
105                ")";
106 } # }}}
107
108 sub delete (@) { #{{{
109         my @files=@_;
110
111         # Remove feed data for removed pages.
112         foreach my $file (@files) {
113                 my $page=IkiWiki::pagename($file);
114                 remove_feeds($page);
115         }
116 } #}}}
117
118 sub loadstate () { #{{{
119         if (-e "$IkiWiki::config{wikistatedir}/aggregate") {
120                 open (IN, "$IkiWiki::config{wikistatedir}/aggregate" ||
121                         die "$IkiWiki::config{wikistatedir}/aggregate: $!");
122                 while (<IN>) {
123                         $_=IkiWiki::possibly_foolish_untaint($_);
124                         chomp;
125                         my $data={};
126                         foreach my $i (split(/ /, $_)) {
127                                 my ($field, $val)=split(/=/, $i, 2);
128                                 if ($field eq "name" || $field eq "feed" ||
129                                     $field eq "guid" || $field eq "message") {
130                                         $data->{$field}=decode_entities($val, " \t\n");
131                                 }
132                                 elsif ($field eq "tag") {
133                                         push @{$data->{tags}}, $val;
134                                 }
135                                 else {
136                                         $data->{$field}=$val;
137                                 }
138                         }
139                         
140                         if (exists $data->{name}) {
141                                 $feeds{$data->{name}}=$data;
142                         }
143                         elsif (exists $data->{guid}) {
144                                 $guids{$data->{guid}}=$data;
145                         }
146                 }
147
148                 close IN;
149         }
150 } #}}}
151
152 sub savestate () { #{{{
153         eval q{use HTML::Entities};
154         die $@ if $@;
155         open (OUT, ">$IkiWiki::config{wikistatedir}/aggregate" ||
156                 die "$IkiWiki::config{wikistatedir}/aggregate: $!");
157         foreach my $data (values %feeds, values %guids) {
158                 if ($data->{remove}) {
159                         if ($data->{name}) {
160                                 foreach my $guid (values %guids) {
161                                         if ($guid->{feed} eq $data->{name}) {
162                                                 $guid->{remove}=1;
163                                         }
164                                 }
165                         }
166                         else {
167                                 unlink pagefile($data->{page});
168                         }
169                         next;
170                 }
171
172                 my @line;
173                 foreach my $field (keys %$data) {
174                         if ($field eq "name" || $field eq "feed" ||
175                             $field eq "guid" || $field eq "message") {
176                                 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
177                         }
178                         elsif ($field eq "tags") {
179                                 push @line, "tag=$_" foreach @{$data->{tags}};
180                         }
181                         else {
182                                 push @line, "$field=".$data->{$field};
183                         }
184                 }
185                 print OUT join(" ", @line)."\n";
186         }
187         close OUT;
188 } #}}}
189
190 sub aggregate () { #{{{
191         eval q{use XML::Feed};
192         die $@ if $@;
193         eval q{use HTML::Entities};
194         die $@ if $@;
195
196         foreach my $feed (values %feeds) {
197                 next unless $IkiWiki::config{rebuild} || 
198                         time - $feed->{lastupdate} >= $feed->{updateinterval};
199                 $feed->{lastupdate}=time;
200                 $feed->{newposts}=0;
201                 $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
202
203                 IkiWiki::debug("checking feed ".$feed->{name}." ...");
204
205                 if (! length $feed->{feedurl}) {
206                         my @urls=XML::Feed->find_feeds($feed->{url});
207                         if (! @urls) {
208                                 $feed->{message}="could not find feed at ".$feed->{feedurl};
209                                 IkiWiki::debug($feed->{message});
210                                 next;
211                         }
212                         $feed->{feedurl}=pop @urls;
213                 }
214                 my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))};
215                 if ($@) {
216                         $feed->{message}="feed crashed XML::Feed! $@";
217                         IkiWiki::debug($feed->{message});
218                         next;
219                 }
220                 if (! $f) {
221                         $feed->{message}=XML::Feed->errstr;
222                         IkiWiki::debug($feed->{message});
223                         next;
224                 }
225
226                 foreach my $entry ($f->entries) {
227                         add_page(
228                                 feed => $feed,
229                                 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
230                                 link => $entry->link,
231                                 content => $entry->content->body,
232                                 guid => defined $entry->id ? $entry->id : time."_".$feed->name,
233                                 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
234                         );
235                 }
236
237                 $feed->{message}="processed ok at ".
238                         IkiWiki::displaytime($feed->{lastupdate});
239         }
240
241         # TODO: expiry
242 } #}}}
243
244 sub add_page (@) { #{{{
245         my %params=@_;
246         
247         my $feed=$params{feed};
248         my $guid={};
249         my $mtime;
250         if (exists $guids{$params{guid}}) {
251                 # updating an existing post
252                 $guid=$guids{$params{guid}};
253         }
254         else {
255                 # new post
256                 $guid->{guid}=$params{guid};
257                 $guids{$params{guid}}=$guid;
258                 $mtime=$params{ctime};
259                 $feed->{numposts}++;
260                 $feed->{newposts}++;
261
262                 # assign it an unused page
263                 my $page=IkiWiki::titlepage($params{title});
264                 # escape slashes and periods in title so it doesn't specify
265                 # directory name or trigger ".." disallowing code.
266                 $page=~s!([/.])!"__".ord($1)."__"!eg;
267                 $page=$feed->{dir}."/".$page;
268                 ($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
269                 if (! defined $page || ! length $page) {
270                         $page=$feed->{dir}."/item";
271                 }
272                 my $c="";
273                 while (exists $IkiWiki::pagecase{lc $page.$c} ||
274                        -e pagefile($page.$c)) {
275                         $c++
276                 }
277                 $guid->{page}=$page;
278                 IkiWiki::debug("creating new page $page");
279         }
280         $guid->{feed}=$feed->{name};
281         
282         # To write or not to write? Need to avoid writing unchanged pages
283         # to avoid unneccessary rebuilding. The mtime from rss cannot be
284         # trusted; let's use a digest.
285         eval q{use Digest::MD5 'md5_hex'};
286         require Encode;
287         my $digest=md5_hex(Encode::encode_utf8($params{content}));
288         return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $IkiWiki::config{rebuild};
289         $guid->{md5}=$digest;
290
291         # Create the page.
292         my $template=IkiWiki::template("aggregatepost.tmpl", blind_cache => 1);
293         $template->param(title => $params{title})
294                 if defined $params{title} && length($params{title});
295         $template->param(content => htmlescape(htmlabs($params{content}, $feed->{feedurl})));
296         $template->param(name => $feed->{name});
297         $template->param(url => $feed->{url});
298         $template->param(permalink => urlabs($params{link}, $feed->{feedurl}))
299                 if defined $params{link};
300         if (ref $feed->{tags}) {
301                 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
302         }
303         IkiWiki::writefile(IkiWiki::htmlpage($guid->{page}), $IkiWiki::config{srcdir},
304                 $template->output);
305
306         # Set the mtime, this lets the build process get the right creation
307         # time on record for the new page.
308         utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
309 } #}}}
310
311 sub htmlescape ($) { #{{{
312         # escape accidental wikilinks and preprocessor stuff
313         my $html=shift;
314         $html=~s/(?<!\\)\[\[/\\\[\[/g;
315         return $html;
316 } #}}}
317
318 sub urlabs ($$) { #{{{
319         my $url=shift;
320         my $urlbase=shift;
321
322         URI->new_abs($url, $urlbase)->as_string;
323 } #}}}
324
325 sub htmlabs ($$) { #{{{
326         # Convert links in html from relative to absolute.
327         # Note that this is a heuristic, which is not specified by the rss
328         # spec and may not be right for all feeds. Also, see Debian
329         # bug #381359.
330         my $html=shift;
331         my $urlbase=shift;
332
333         my $ret="";
334         my $p = HTML::Parser->new(api_version => 3);
335         $p->handler(default => sub { $ret.=join("", @_) }, "text");
336         $p->handler(start => sub {
337                 my ($tagname, $pos, $text) = @_;
338                 if (ref $HTML::Tagset::linkElements{$tagname}) {
339                         while (4 <= @$pos) {
340                                 # use attribute sets from right to left
341                                 # to avoid invalidating the offsets
342                                 # when replacing the values
343                                 my($k_offset, $k_len, $v_offset, $v_len) =
344                                         splice(@$pos, -4);
345                                 my $attrname = lc(substr($text, $k_offset, $k_len));
346                                 next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
347                                 next unless $v_offset; # 0 v_offset means no value
348                                 my $v = substr($text, $v_offset, $v_len);
349                                 $v =~ s/^([\'\"])(.*)\1$/$2/;
350                                 my $new_v=urlabs($v, $urlbase);
351                                 $new_v =~ s/\"/&quot;/g; # since we quote with ""
352                                 substr($text, $v_offset, $v_len) = qq("$new_v");
353                         }
354                 }
355                 $ret.=$text;
356         }, "tagname, tokenpos, text");
357         $p->parse($html);
358         $p->eof;
359
360         return $ret;
361 } #}}}
362
363 sub remove_feeds () { #{{{
364         my $page=shift;
365
366         my %removed;
367         foreach my $id (keys %feeds) {
368                 if ($feeds{$id}->{sourcepage} eq $page) {
369                         $feeds{$id}->{remove}=1;
370                         $removed{$id}=1;
371                 }
372         }
373 } #}}}
374
375 sub pagefile ($) { #{{{
376         my $page=shift;
377
378         return "$IkiWiki::config{srcdir}/".IkiWiki::htmlpage($page);
379 } #}}}
380
381 1