1a9f94a124d766687da1daeac391cf8b479a4cec
[ikiwiki.git] / IkiWiki / Plugin / meta.pm
1 #!/usr/bin/perl
2 # Ikiwiki metadata plugin.
3 package IkiWiki::Plugin::meta;
4
5 use warnings;
6 use strict;
7 use IkiWiki 3.00;
8
9 my %metaheaders;
10
11 sub import {
12         hook(type => "getsetup", id => "meta", call => \&getsetup);
13         hook(type => "needsbuild", id => "meta", call => \&needsbuild);
14         hook(type => "preprocess", id => "meta", call => \&preprocess, scan => 1);
15         hook(type => "pagetemplate", id => "meta", call => \&pagetemplate);
16 }
17
18 sub getsetup () {
19         return
20                 plugin => {
21                         safe => 1,
22                         rebuild => undef,
23                         section => "core",
24                 },
25 }
26
27 sub needsbuild (@) {
28         my $needsbuild=shift;
29         foreach my $page (keys %pagestate) {
30                 if (exists $pagestate{$page}{meta}) {
31                         if (exists $pagesources{$page} &&
32                             grep { $_ eq $pagesources{$page} } @$needsbuild) {
33                                 # remove state, it will be re-added
34                                 # if the preprocessor directive is still
35                                 # there during the rebuild
36                                 delete $pagestate{$page}{meta};
37                         }
38                 }
39         }
40         return $needsbuild;
41 }
42
43 sub scrub ($$$) {
44         if (IkiWiki::Plugin::htmlscrubber->can("sanitize")) {
45                 return IkiWiki::Plugin::htmlscrubber::sanitize(
46                         content => shift, page => shift, destpage => shift);
47         }
48         else {
49                 return shift;
50         }
51 }
52
53 sub safeurl ($) {
54         my $url=shift;
55         if (exists $IkiWiki::Plugin::htmlscrubber::{safe_url_regexp} &&
56             defined $IkiWiki::Plugin::htmlscrubber::safe_url_regexp) {
57                 return $url=~/$IkiWiki::Plugin::htmlscrubber::safe_url_regexp/;
58         }
59         else {
60                 return 1;
61         }
62 }
63
64 sub htmlize ($$$) {
65         my $page = shift;
66         my $destpage = shift;
67
68         return IkiWiki::htmlize($page, $destpage, pagetype($pagesources{$page}),
69                 IkiWiki::linkify($page, $destpage,
70                 IkiWiki::preprocess($page, $destpage, shift)));
71 }
72
73 sub preprocess (@) {
74         return "" unless @_;
75         my %params=@_;
76         my $key=shift;
77         my $value=$params{$key};
78         delete $params{$key};
79         my $page=$params{page};
80         delete $params{page};
81         my $destpage=$params{destpage};
82         delete $params{destpage};
83         delete $params{preview};
84
85         eval q{use HTML::Entities};
86         # Always decode, even if encoding later, since it might not be
87         # fully encoded.
88         $value=decode_entities($value);
89
90         # Metadata collection that needs to happen during the scan pass.
91         if ($key eq 'title') {
92                 $pagestate{$page}{meta}{title}=$value;
93                 if (exists $params{sortas}) {
94                         $pagestate{$page}{meta}{titlesort}=$params{sortas};
95                 }
96                 else {
97                         delete $pagestate{$page}{meta}{titlesort};
98                 }
99                 return "";
100         }
101         elsif ($key eq 'description') {
102                 $pagestate{$page}{meta}{description}=$value;
103                 # fallthrough
104         }
105         elsif ($key eq 'guid') {
106                 $pagestate{$page}{meta}{guid}=$value;
107                 # fallthrough
108         }
109         elsif ($key eq 'license') {
110                 push @{$metaheaders{$page}}, '<link rel="license" href="#page_license" />';
111                 $pagestate{$page}{meta}{license}=$value;
112                 return "";
113         }
114         elsif ($key eq 'copyright') {
115                 push @{$metaheaders{$page}}, '<link rel="copyright" href="#page_copyright" />';
116                 $pagestate{$page}{meta}{copyright}=$value;
117                 return "";
118         }
119         elsif ($key eq 'link' && ! %params) {
120                 # hidden WikiLink
121                 add_link($page, $value);
122                 return "";
123         }
124         elsif ($key eq 'author') {
125                 $pagestate{$page}{meta}{author}=$value;
126                 if (exists $params{sortas}) {
127                         $pagestate{$page}{meta}{authorsort}=$params{sortas};
128                 }
129                 else {
130                         delete $pagestate{$page}{meta}{authorsort};
131                 }
132                 # fallthorough
133         }
134         elsif ($key eq 'authorurl') {
135                 $pagestate{$page}{meta}{authorurl}=$value if safeurl($value);
136                 # fallthrough
137         }
138         elsif ($key eq 'permalink') {
139                 $pagestate{$page}{meta}{permalink}=$value if safeurl($value);
140                 # fallthrough
141         }
142         elsif ($key eq 'date') {
143                 eval q{use Date::Parse};
144                 if (! $@) {
145                         my $time = str2time($value);
146                         $IkiWiki::pagectime{$page}=$time if defined $time;
147                 }
148         }
149         elsif ($key eq 'updated') {
150                 eval q{use Date::Parse};
151                 if (! $@) {
152                         my $time = str2time($value);
153                         $pagestate{$page}{meta}{updated}=$time if defined $time;
154                 }
155         }
156
157         if (! defined wantarray) {
158                 # avoid collecting duplicate data during scan pass
159                 return;
160         }
161
162         # Metadata handling that happens only during preprocessing pass.
163         if ($key eq 'permalink') {
164                 if (safeurl($value)) {
165                         push @{$metaheaders{$page}}, scrub('<link rel="bookmark" href="'.encode_entities($value).'" />', $page, $destpage);
166                 }
167         }
168         elsif ($key eq 'stylesheet') {
169                 my $rel=exists $params{rel} ? $params{rel} : "alternate stylesheet";
170                 my $title=exists $params{title} ? $params{title} : $value;
171                 # adding .css to the value prevents using any old web
172                 # editable page as a stylesheet
173                 my $stylesheet=bestlink($page, $value.".css");
174                 if (! length $stylesheet) {
175                         error gettext("stylesheet not found")
176                 }
177                 push @{$metaheaders{$page}}, scrub('<link href="'.urlto($stylesheet, $page).
178                         '" rel="'.encode_entities($rel).
179                         '" title="'.encode_entities($title).
180                         "\" type=\"text/css\" />", $page, $destpage);
181         }
182         elsif ($key eq 'openid') {
183                 my $delegate=0; # both by default
184                 if (exists $params{delegate}) {
185                         $delegate = 1 if lc $params{delegate} eq 'openid';
186                         $delegate = 2 if lc $params{delegate} eq 'openid2';
187                 }
188                 if (exists $params{server} && safeurl($params{server})) {
189                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
190                                 '" rel="openid.server" />' if $delegate ne 2;
191                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
192                                 '" rel="openid2.provider" />' if $delegate ne 1;
193                 }
194                 if (safeurl($value)) {
195                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
196                                 '" rel="openid.delegate" />' if $delegate ne 2;
197                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
198                                 '" rel="openid2.local_id" />' if $delegate ne 1;
199                 }
200                 if (exists $params{"xrds-location"} && safeurl($params{"xrds-location"})) {
201                         # force url absolute
202                         eval q{use URI};
203                         error($@) if $@;
204                         my $url=URI->new_abs($params{"xrds-location"}, $config{url});
205                         push @{$metaheaders{$page}}, '<meta http-equiv="X-XRDS-Location" '.
206                                 'content="'.encode_entities($url).'" />';
207                 }
208         }
209         elsif ($key eq 'redir') {
210                 return "" if $page ne $destpage;
211                 my $safe=0;
212                 if ($value !~ /^\w+:\/\//) {
213                         my ($redir_page, $redir_anchor) = split /\#/, $value;
214
215                         my $link=bestlink($page, $redir_page);
216                         if (! length $link) {
217                                 error gettext("redir page not found")
218                         }
219                         add_depends($page, $link, deptype("presence"));
220
221                         $value=urlto($link, $page);
222                         $value.='#'.$redir_anchor if defined $redir_anchor;
223                         $safe=1;
224
225                         # redir cycle detection
226                         $pagestate{$page}{meta}{redir}=$link;
227                         my $at=$page;
228                         my %seen;
229                         while (exists $pagestate{$at}{meta}{redir}) {
230                                 if ($seen{$at}) {
231                                         error gettext("redir cycle is not allowed")
232                                 }
233                                 $seen{$at}=1;
234                                 $at=$pagestate{$at}{meta}{redir};
235                         }
236                 }
237                 else {
238                         $value=encode_entities($value);
239                 }
240                 my $delay=int(exists $params{delay} ? $params{delay} : 0);
241                 my $redir="<meta http-equiv=\"refresh\" content=\"$delay; URL=$value\" />";
242                 if (! $safe) {
243                         $redir=scrub($redir, $page, $destpage);
244                 }
245                 push @{$metaheaders{$page}}, $redir;
246         }
247         elsif ($key eq 'link') {
248                 if (%params) {
249                         push @{$metaheaders{$page}}, scrub("<link href=\"".encode_entities($value)."\" ".
250                                 join(" ", map {
251                                         encode_entities($_)."=\"".encode_entities(decode_entities($params{$_}))."\""
252                                 } keys %params).
253                                 " />\n", $page, $destpage);
254                 }
255         }
256         elsif ($key eq 'robots') {
257                 push @{$metaheaders{$page}}, '<meta name="robots"'.
258                         ' content="'.encode_entities($value).'" />';
259         }
260         elsif ($key eq 'description') {
261                 push @{$metaheaders{$page}}, '<meta name="'.
262                         encode_entities($key).
263                         '" content="'.encode_entities($value).'" />';
264         }
265         elsif ($key eq 'name') {
266                 push @{$metaheaders{$page}}, scrub('<meta '.$key.'="'.
267                         encode_entities($value).
268                         join(' ', map { "$_=\"$params{$_}\"" } keys %params).
269                         ' />', $page, $destpage);
270         }
271         else {
272                 push @{$metaheaders{$page}}, scrub('<meta name="'.
273                         encode_entities($key).'" content="'.
274                         encode_entities($value).'" />', $page, $destpage);
275         }
276
277         return "";
278 }
279
280 sub pagetemplate (@) {
281         my %params=@_;
282         my $page=$params{page};
283         my $destpage=$params{destpage};
284         my $template=$params{template};
285
286         if (exists $metaheaders{$page} && $template->query(name => "meta")) {
287                 # avoid duplicate meta lines
288                 my %seen;
289                 $template->param(meta => join("\n", grep { (! $seen{$_}) && ($seen{$_}=1) } @{$metaheaders{$page}}));
290         }
291         if (exists $pagestate{$page}{meta}{title} && $template->query(name => "title")) {
292                 $template->param(title => HTML::Entities::encode_numeric($pagestate{$page}{meta}{title}));
293                 $template->param(title_overridden => 1);
294         }
295
296         foreach my $field (qw{author authorurl permalink}) {
297                 $template->param($field => $pagestate{$page}{meta}{$field})
298                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
299         }
300
301         foreach my $field (qw{permalink}) {
302                 $template->param($field => IkiWiki::urlabs($pagestate{$page}{meta}{$field}, $config{url}))
303                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
304         }
305
306         foreach my $field (qw{description}) {
307                 $template->param($field => HTML::Entities::encode_numeric($pagestate{$page}{meta}{$field}))
308                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
309         }
310
311         foreach my $field (qw{license copyright}) {
312                 if (exists $pagestate{$page}{meta}{$field} && $template->query(name => $field) &&
313                     ($page eq $destpage || ! exists $pagestate{$destpage}{meta}{$field} ||
314                      $pagestate{$page}{meta}{$field} ne $pagestate{$destpage}{meta}{$field})) {
315                         $template->param($field => htmlize($page, $destpage, $pagestate{$page}{meta}{$field}));
316                 }
317         }
318 }
319
320 sub get_sort_key {
321         my $page = shift;
322         my $meta = shift;
323
324         # e.g. titlesort (also makes sense for author)
325         my $key = $pagestate{$page}{meta}{$meta . "sort"};
326         return $key if defined $key;
327
328         # e.g. title
329         $key = $pagestate{$page}{meta}{$meta};
330         return $key if defined $key;
331
332         # fall back to closer-to-core things
333         if ($meta eq 'title') {
334                 return pagetitle(IkiWiki::basename($page));
335         }
336         elsif ($meta eq 'date') {
337                 return $IkiWiki::pagectime{$page};
338         }
339         elsif ($meta eq 'updated') {
340                 return $IkiWiki::pagemtime{$page};
341         }
342         else {
343                 return '';
344         }
345 }
346
347 sub match {
348         my $field=shift;
349         my $page=shift;
350         
351         # turn glob into a safe regexp
352         my $re=IkiWiki::glob2re(shift);
353
354         my $val;
355         if (exists $pagestate{$page}{meta}{$field}) {
356                 $val=$pagestate{$page}{meta}{$field};
357         }
358         elsif ($field eq 'title') {
359                 $val = pagetitle($page);
360         }
361
362         if (defined $val) {
363                 if ($val=~$re) {
364                         return IkiWiki::SuccessReason->new("$re matches $field of $page", $page => $IkiWiki::DEPEND_CONTENT, "" => 1);
365                 }
366                 else {
367                         return IkiWiki::FailReason->new("$re does not match $field of $page", $page => $IkiWiki::DEPEND_CONTENT, "" => 1);
368                 }
369         }
370         else {
371                 return IkiWiki::FailReason->new("$page does not have a $field", $page => $IkiWiki::DEPEND_CONTENT);
372         }
373 }
374
375 package IkiWiki::PageSpec;
376
377 sub match_title ($$;@) {
378         IkiWiki::Plugin::meta::match("title", @_);
379 }
380
381 sub match_author ($$;@) {
382         IkiWiki::Plugin::meta::match("author", @_);
383 }
384
385 sub match_authorurl ($$;@) {
386         IkiWiki::Plugin::meta::match("authorurl", @_);
387 }
388
389 sub match_license ($$;@) {
390         IkiWiki::Plugin::meta::match("license", @_);
391 }
392
393 sub match_copyright ($$;@) {
394         IkiWiki::Plugin::meta::match("copyright", @_);
395 }
396
397 sub match_guid ($$;@) {
398         IkiWiki::Plugin::meta::match("guid", @_);
399 }
400
401 package IkiWiki::SortSpec;
402
403 sub cmp_meta {
404         my $meta = shift;
405         error(gettext("sort=meta requires a parameter")) unless defined $meta;
406
407         if ($meta eq 'updated' || $meta eq 'date') {
408                 return IkiWiki::Plugin::meta::get_sort_key($a, $meta)
409                         <=>
410                         IkiWiki::Plugin::meta::get_sort_key($b, $meta);
411         }
412
413         return IkiWiki::Plugin::meta::get_sort_key($a, $meta)
414                 cmp
415                 IkiWiki::Plugin::meta::get_sort_key($b, $meta);
416 }
417
418 # A prototype of how sort=title could behave in 4.0 or something
419 sub cmp_meta_title {
420         $_[0] = 'title';
421         return cmp_meta(@_);
422 }
423
424 1