]> git.vanrenterghem.biz Git - git.ikiwiki.info.git/blob - IkiWiki/Plugin/meta.pm
Merge branch 'master' of ssh://git.ikiwiki.info
[git.ikiwiki.info.git] / IkiWiki / Plugin / meta.pm
1 #!/usr/bin/perl
2 # Ikiwiki metadata plugin.
3 package IkiWiki::Plugin::meta;
5 use warnings;
6 use strict;
7 use IkiWiki 3.00;
9 my %metaheaders;
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 }
18 sub getsetup () {
19         return
20                 plugin => {
21                         safe => 1,
22                         rebuild => undef,
23                         section => "core",
24                 },
25 }
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 }
43 sub scrub ($$$) {
44         if (IkiWiki::Plugin::htmlscrubber->can("sanitize")) {
45                 return IkiWiki::Plugin::htmlscrubber::sanitize(
46                         content => shift, page => shift, destpage => shift);
47         }
48         else {
49                 return shift;
50         }
51 }
53 sub safeurl ($) {
54         my $url=shift;
55         if (exists $IkiWiki::Plugin::htmlscrubber::{safe_url_regexp} &&
56             defined $IkiWiki::Plugin::htmlscrubber::safe_url_regexp) {
57                 return $url=~/$IkiWiki::Plugin::htmlscrubber::safe_url_regexp/;
58         }
59         else {
60                 return 1;
61         }
62 }
64 sub htmlize ($$$) {
65         my $page = shift;
66         my $destpage = shift;
68         return IkiWiki::htmlize($page, $destpage, pagetype($pagesources{$page}),
69                 IkiWiki::linkify($page, $destpage,
70                 IkiWiki::preprocess($page, $destpage, shift)));
71 }
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};
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);
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="#pagelicense" />';
111                 $pagestate{$page}{meta}{license}=$value;
112                 return "";
113         }
114         elsif ($key eq 'copyright') {
115                 push @{$metaheaders{$page}}, '<link rel="copyright" href="#pagecopyright" />';
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 'enclosure') {
125                 my $link=bestlink($page, $value);
126                 if (! length $link) {
127                         error gettext("enclosure not found")
128                 }
129                 add_depends($page, $link, deptype("presence"));
131                 $value=urlto($link, $page, 1);
132                 $pagestate{$page}{meta}{enclosure}=$value;
133                 $pagestate{$page}{meta}{enclosurefile}=$link;
134                 # fallthrough
135         }
136         elsif ($key eq 'author') {
137                 $pagestate{$page}{meta}{author}=$value;
138                 if (exists $params{sortas}) {
139                         $pagestate{$page}{meta}{authorsort}=$params{sortas};
140                 }
141                 else {
142                         delete $pagestate{$page}{meta}{authorsort};
143                 }
144                 # fallthorough
145         }
146         elsif ($key eq 'authorurl') {
147                 $pagestate{$page}{meta}{authorurl}=$value if safeurl($value);
148                 # fallthrough
149         }
150         elsif ($key eq 'permalink') {
151                 $pagestate{$page}{meta}{permalink}=$value if safeurl($value);
152                 # fallthrough
153         }
154         elsif ($key eq 'date') {
155                 eval q{use Date::Parse};
156                 if (! $@) {
157                         my $time = str2time($value);
158                         $IkiWiki::pagectime{$page}=$time if defined $time;
159                 }
160         }
161         elsif ($key eq 'updated') {
162                 eval q{use Date::Parse};
163                 if (! $@) {
164                         my $time = str2time($value);
165                         $pagestate{$page}{meta}{updated}=$time if defined $time;
166                 }
167         }
169         if (! defined wantarray) {
170                 # avoid collecting duplicate data during scan pass
171                 return;
172         }
174         # Metadata handling that happens only during preprocessing pass.
175         if ($key eq 'permalink') {
176                 if (safeurl($value)) {
177                         push @{$metaheaders{$page}}, scrub('<link rel="bookmark" href="'.encode_entities($value).'" />', $page, $destpage);
178                 }
179         }
180         elsif ($key eq 'stylesheet') {
181                 my $rel=exists $params{rel} ? $params{rel} : "alternate stylesheet";
182                 my $title=exists $params{title} ? $params{title} : $value;
183                 # adding .css to the value prevents using any old web
184                 # editable page as a stylesheet
185                 my $stylesheet=bestlink($page, $value.".css");
186                 if (! length $stylesheet) {
187                         error gettext("stylesheet not found")
188                 }
189                 push @{$metaheaders{$page}}, scrub('<link href="'.urlto($stylesheet, $page).
190                         '" rel="'.encode_entities($rel).
191                         '" title="'.encode_entities($title).
192                         "\" type=\"text/css\" />", $page, $destpage);
193         }
194         elsif ($key eq 'script') {
195                 my $defer=exists $params{defer} ? ' defer="defer"' : '';
196                 my $async=exists $params{async} ? ' async="async"' : '';
197                 my $js=bestlink($page, $value.".js");
198                 if (! length $js) {
199                         error gettext("script not found");
200                 }
201                 push @{$metaheaders{$page}}, scrub('<script src="'.urlto($js, $page).
202                         '"' . $defer . $async . ' type="text/javascript"></script>',
203                         $page, $destpage);
204         }
205         elsif ($key eq 'openid') {
206                 my $delegate=0; # both by default
207                 if (exists $params{delegate}) {
208                         $delegate = 1 if lc $params{delegate} eq 'openid';
209                         $delegate = 2 if lc $params{delegate} eq 'openid2';
210                 }
211                 if (exists $params{server} && safeurl($params{server})) {
212                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
213                                 '" rel="openid.server" />' if $delegate ne 2;
214                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
215                                 '" rel="openid2.provider" />' if $delegate ne 1;
216                 }
217                 if (safeurl($value)) {
218                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
219                                 '" rel="openid.delegate" />' if $delegate ne 2;
220                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
221                                 '" rel="openid2.local_id" />' if $delegate ne 1;
222                 }
223                 if (exists $params{"xrds-location"} && safeurl($params{"xrds-location"})) {
224                         # force url absolute
225                         eval q{use URI};
226                         error($@) if $@;
227                         my $url=URI->new_abs($params{"xrds-location"}, $config{url});
228                         push @{$metaheaders{$page}}, '<meta http-equiv="X-XRDS-Location" '.
229                                 'content="'.encode_entities($url).'" />';
230                 }
231         }
232         elsif ($key eq 'foaf') {
233                 if (safeurl($value)) {
234                         push @{$metaheaders{$page}}, '<link rel="meta" '.
235                                 'type="application/rdf+xml" title="FOAF" '.
236                                 'href="'.encode_entities($value).'" />';
237                 }
238         }
239         elsif ($key eq 'redir') {
240                 return "" if $page ne $destpage;
241                 my $safe=0;
242                 if ($value !~ /^\w+:\/\//) {
243                         my ($redir_page, $redir_anchor) = split /\#/, $value;
245                         my $link=bestlink($page, $redir_page);
246                         if (! length $link) {
247                                 error gettext("redir page not found")
248                         }
249                         add_depends($page, $link, deptype("presence"));
251                         $value=urlto($link, $page);
252                         $value.='#'.$redir_anchor if defined $redir_anchor;
253                         $safe=1;
255                         # redir cycle detection
256                         $pagestate{$page}{meta}{redir}=$link;
257                         my $at=$page;
258                         my %seen;
259                         while (exists $pagestate{$at}{meta}{redir}) {
260                                 if ($seen{$at}) {
261                                         error gettext("redir cycle is not allowed")
262                                 }
263                                 $seen{$at}=1;
264                                 $at=$pagestate{$at}{meta}{redir};
265                         }
266                 }
267                 else {
268                         $value=encode_entities($value);
269                 }
270                 my $delay=int(exists $params{delay} ? $params{delay} : 0);
271                 my $redir="<meta http-equiv=\"refresh\" content=\"$delay; URL=$value\" />";
272                 if (! $safe) {
273                         $redir=scrub($redir, $page, $destpage);
274                 }
275                 push @{$metaheaders{$page}}, $redir;
276         }
277         elsif ($key eq 'link') {
278                 if (%params) {
279                         push @{$metaheaders{$page}}, scrub("<link href=\"".encode_entities($value)."\" ".
280                                 join(" ", map {
281                                         encode_entities($_)."=\"".encode_entities(decode_entities($params{$_}))."\""
282                                 } keys %params).
283                                 " />\n", $page, $destpage);
284                 }
285         }
286         elsif ($key eq 'robots') {
287                 push @{$metaheaders{$page}}, '<meta name="robots"'.
288                         ' content="'.encode_entities($value).'" />';
289         }
290         elsif ($key eq 'description' || $key eq 'author') {
291                 push @{$metaheaders{$page}}, '<meta name="'.$key.
292                         '" content="'.encode_entities($value).'" />';
293         }
294         elsif ($key eq 'name') {
295                 push @{$metaheaders{$page}}, scrub('<meta name="'.
296                         encode_entities($value).
297                         join(' ', map { "$_=\"$params{$_}\"" } keys %params).
298                         ' />', $page, $destpage);
299         }
300         elsif ($key eq 'keywords') {
301                 # Make sure the keyword string is safe: only allow alphanumeric
302                 # characters, space and comma and strip the rest.
303                 $value =~ s/[^[:alnum:], ]+//g;
304                 push @{$metaheaders{$page}}, '<meta name="keywords"'.
305                         ' content="'.encode_entities($value).'" />';
306         }
307         else {
308                 push @{$metaheaders{$page}}, scrub('<meta name="'.
309                         encode_entities($key).'" content="'.
310                         encode_entities($value).'" />', $page, $destpage);
311         }
313         return "";
316 sub pagetemplate (@) {
317         my %params=@_;
318         my $page=$params{page};
319         my $destpage=$params{destpage};
320         my $template=$params{template};
322         if (exists $metaheaders{$page} && $template->query(name => "meta")) {
323                 # avoid duplicate meta lines
324                 my %seen;
325                 $template->param(meta => join("\n", grep { (! $seen{$_}) && ($seen{$_}=1) } @{$metaheaders{$page}}));
326         }
327         if (exists $pagestate{$page}{meta}{title} && $template->query(name => "title")) {
328                 eval q{use HTML::Entities};
329                 $template->param(title => HTML::Entities::encode_numeric($pagestate{$page}{meta}{title}));
330                 $template->param(title_overridden => 1);
331         }
333         if (exists $pagestate{$page}{meta}{enclosure}) {
334                 $template->param(enclosure => HTML::Entities::encode_entities(IkiWiki::urlabs($pagestate{$page}{meta}{enclosure}, $config{url})));
335         }
337         foreach my $field (qw{authorurl}) {
338                 eval q{use HTML::Entities};
339                 $template->param($field => HTML::Entities::encode_entities($pagestate{$page}{meta}{$field}))
340                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
341         }
343         foreach my $field (qw{permalink}) {
344                 if (exists $pagestate{$page}{meta}{$field} && $template->query(name => $field)) {
345                         eval q{use HTML::Entities};
346                         $template->param($field => HTML::Entities::encode_entities(IkiWiki::urlabs($pagestate{$page}{meta}{$field}, $config{url})));
347                 }
348         }
350         foreach my $field (qw{description author}) {
351                 eval q{use HTML::Entities};
352                 $template->param($field => HTML::Entities::encode_numeric($pagestate{$page}{meta}{$field}))
353                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
354         }
356         foreach my $field (qw{license copyright}) {
357                 if (exists $pagestate{$page}{meta}{$field} && $template->query(name => $field) &&
358                     ($page eq $destpage || ! exists $pagestate{$destpage}{meta}{$field} ||
359                      $pagestate{$page}{meta}{$field} ne $pagestate{$destpage}{meta}{$field})) {
360                         $template->param($field => htmlize($page, $destpage, $pagestate{$page}{meta}{$field}));
361                 }
362         }
365 sub get_sort_key {
366         my $page = shift;
367         my $meta = shift;
369         # e.g. titlesort (also makes sense for author)
370         my $key = $pagestate{$page}{meta}{$meta . "sort"};
371         return $key if defined $key;
373         # e.g. title
374         $key = $pagestate{$page}{meta}{$meta};
375         return $key if defined $key;
377         # fall back to closer-to-core things
378         if ($meta eq 'title') {
379                 return pagetitle(IkiWiki::basename($page));
380         }
381         elsif ($meta eq 'date') {
382                 return $IkiWiki::pagectime{$page};
383         }
384         elsif ($meta eq 'updated') {
385                 return $IkiWiki::pagemtime{$page};
386         }
387         else {
388                 return '';
389         }
392 sub match {
393         my $field=shift;
394         my $page=shift;
395         
396         # turn glob into a safe regexp
397         my $re=IkiWiki::glob2re(shift);
399         my $val;
400         if (exists $pagestate{$page}{meta}{$field}) {
401                 $val=$pagestate{$page}{meta}{$field};
402         }
403         elsif ($field eq 'title') {
404                 $val = pagetitle($page);
405         }
407         if (defined $val) {
408                 if ($val=~$re) {
409                         return IkiWiki::SuccessReason->new("$re matches $field of $page", $page => $IkiWiki::DEPEND_CONTENT, "" => 1);
410                 }
411                 else {
412                         return IkiWiki::FailReason->new("$re does not match $field of $page", $page => $IkiWiki::DEPEND_CONTENT, "" => 1);
413                 }
414         }
415         else {
416                 return IkiWiki::FailReason->new("$page does not have a $field", $page => $IkiWiki::DEPEND_CONTENT);
417         }
420 package IkiWiki::PageSpec;
422 sub match_title ($$;@) {
423         IkiWiki::Plugin::meta::match("title", @_);
426 sub match_author ($$;@) {
427         IkiWiki::Plugin::meta::match("author", @_);
430 sub match_authorurl ($$;@) {
431         IkiWiki::Plugin::meta::match("authorurl", @_);
434 sub match_license ($$;@) {
435         IkiWiki::Plugin::meta::match("license", @_);
438 sub match_copyright ($$;@) {
439         IkiWiki::Plugin::meta::match("copyright", @_);
442 sub match_guid ($$;@) {
443         IkiWiki::Plugin::meta::match("guid", @_);
446 package IkiWiki::SortSpec;
448 sub cmp_meta {
449         my $meta = shift;
450         error(gettext("sort=meta requires a parameter")) unless defined $meta;
452         if ($meta eq 'updated' || $meta eq 'date') {
453                 return IkiWiki::Plugin::meta::get_sort_key($a, $meta)
454                         <=>
455                         IkiWiki::Plugin::meta::get_sort_key($b, $meta);
456         }
458         return IkiWiki::Plugin::meta::get_sort_key($a, $meta)
459                 cmp
460                 IkiWiki::Plugin::meta::get_sort_key($b, $meta);
463 # A prototype of how sort=title could behave in 4.0 or something
464 sub cmp_meta_title {
465         $_[0] = 'title';
466         return cmp_meta(@_);