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 'script') {
 183                 my $defer=exists $params{defer} ? ' defer="defer"' : '';
 184                 my $async=exists $params{async} ? ' async="async"' : '';
 185                 my $js=bestlink($page, $value.".js");
 186                 if (! length $js) {
 187                         error gettext("script not found");
 188                 }
 189                 push @{$metaheaders{$page}}, scrub('<script src="'.urlto($js, $page).
 190                         '"' . $defer . $async . ' type="text/javascript"></script>',
 191                         $page, $destpage);
 192         }
 193         elsif ($key eq 'openid') {
 194                 my $delegate=0; # both by default
 195                 if (exists $params{delegate}) {
 196                         $delegate = 1 if lc $params{delegate} eq 'openid';
 197                         $delegate = 2 if lc $params{delegate} eq 'openid2';
 198                 }
 199                 if (exists $params{server} && safeurl($params{server})) {
 200                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
 201                                 '" rel="openid.server" />' if $delegate ne 2;
 202                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
 203                                 '" rel="openid2.provider" />' if $delegate ne 1;
 204                 }
 205                 if (safeurl($value)) {
 206                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
 207                                 '" rel="openid.delegate" />' if $delegate ne 2;
 208                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
 209                                 '" rel="openid2.local_id" />' if $delegate ne 1;
 210                 }
 211                 if (exists $params{"xrds-location"} && safeurl($params{"xrds-location"})) {
 212                         # force url absolute
 213                         eval q{use URI};
 214                         error($@) if $@;
 215                         my $url=URI->new_abs($params{"xrds-location"}, $config{url});
 216                         push @{$metaheaders{$page}}, '<meta http-equiv="X-XRDS-Location" '.
 217                                 'content="'.encode_entities($url).'" />';
 218                 }
 219         }
 220         elsif ($key eq 'foaf') {
 221                 if (safeurl($value)) {
 222                         push @{$metaheaders{$page}}, '<link rel="meta" '.
 223                                 'type="application/rdf+xml" title="FOAF" '.
 224                                 'href="'.encode_entities($value).'" />';
 225                 }
 226         }
 227         elsif ($key eq 'redir') {
 228                 return "" if $page ne $destpage;
 229                 my $safe=0;
 230                 if ($value !~ /^\w+:\/\//) {
 231                         my ($redir_page, $redir_anchor) = split /\#/, $value;
 232
 233                         my $link=bestlink($page, $redir_page);
 234                         if (! length $link) {
 235                                 error gettext("redir page not found")
 236                         }
 237                         add_depends($page, $link, deptype("presence"));
 238
 239                         $value=urlto($link, $page);
 240                         $value.='#'.$redir_anchor if defined $redir_anchor;
 241                         $safe=1;
 242
 243                         # redir cycle detection
 244                         $pagestate{$page}{meta}{redir}=$link;
 245                         my $at=$page;
 246                         my %seen;
 247                         while (exists $pagestate{$at}{meta}{redir}) {
 248                                 if ($seen{$at}) {
 249                                         error gettext("redir cycle is not allowed")
 250                                 }
 251                                 $seen{$at}=1;
 252                                 $at=$pagestate{$at}{meta}{redir};
 253                         }
 254                 }
 255                 else {
 256                         $value=encode_entities($value);
 257                 }
 258                 my $delay=int(exists $params{delay} ? $params{delay} : 0);
 259                 my $redir="<meta http-equiv=\"refresh\" content=\"$delay; URL=$value\" />";
 260                 if (! $safe) {
 261                         $redir=scrub($redir, $page, $destpage);
 262                 }
 263                 push @{$metaheaders{$page}}, $redir;
 264         }
 265         elsif ($key eq 'link') {
 266                 if (%params) {
 267                         push @{$metaheaders{$page}}, scrub("<link href=\"".encode_entities($value)."\" ".
 268                                 join(" ", map {
 269                                         encode_entities($_)."=\"".encode_entities(decode_entities($params{$_}))."\""
 270                                 } keys %params).
 271                                 " />\n", $page, $destpage);
 272                 }
 273         }
 274         elsif ($key eq 'robots') {
 275                 push @{$metaheaders{$page}}, '<meta name="robots"'.
 276                         ' content="'.encode_entities($value).'" />';
 277         }
 278         elsif ($key eq 'description') {
 279                 push @{$metaheaders{$page}}, '<meta name="'.
 280                         encode_entities($key).
 281                         '" content="'.encode_entities($value).'" />';
 282         }
 283         elsif ($key eq 'name') {
 284                 push @{$metaheaders{$page}}, scrub('<meta '.$key.'="'.
 285                         encode_entities($value).
 286                         join(' ', map { "$_=\"$params{$_}\"" } keys %params).
 287                         ' />', $page, $destpage);
 288         }
 289         else {
 290                 push @{$metaheaders{$page}}, scrub('<meta name="'.
 291                         encode_entities($key).'" content="'.
 292                         encode_entities($value).'" />', $page, $destpage);
 293         }
 294
 295         return "";
 296 }
 297
 298 sub pagetemplate (@) {
 299         my %params=@_;
 300         my $page=$params{page};
 301         my $destpage=$params{destpage};
 302         my $template=$params{template};
 303
 304         if (exists $metaheaders{$page} && $template->query(name => "meta")) {
 305                 # avoid duplicate meta lines
 306                 my %seen;
 307                 $template->param(meta => join("\n", grep { (! $seen{$_}) && ($seen{$_}=1) } @{$metaheaders{$page}}));
 308         }
 309         if (exists $pagestate{$page}{meta}{title} && $template->query(name => "title")) {
 310                 eval q{use HTML::Entities};
 311                 $template->param(title => HTML::Entities::encode_numeric($pagestate{$page}{meta}{title}));
 312                 $template->param(title_overridden => 1);
 313         }
 314
 315         foreach my $field (qw{author authorurl permalink}) {
 316                 $template->param($field => $pagestate{$page}{meta}{$field})
 317                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
 318         }
 319
 320         foreach my $field (qw{permalink}) {
 321                 $template->param($field => IkiWiki::urlabs($pagestate{$page}{meta}{$field}, $config{url}))
 322                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
 323         }
 324
 325         foreach my $field (qw{description}) {
 326                 eval q{use HTML::Entities};
 327                 $template->param($field => HTML::Entities::encode_numeric($pagestate{$page}{meta}{$field}))
 328                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
 329         }
 330
 331         foreach my $field (qw{license copyright}) {
 332                 if (exists $pagestate{$page}{meta}{$field} && $template->query(name => $field) &&
 333                     ($page eq $destpage || ! exists $pagestate{$destpage}{meta}{$field} ||
 334                      $pagestate{$page}{meta}{$field} ne $pagestate{$destpage}{meta}{$field})) {
 335                         $template->param($field => htmlize($page, $destpage, $pagestate{$page}{meta}{$field}));
 336                 }
 337         }
 338 }
 339
 340 sub get_sort_key {
 341         my $page = shift;
 342         my $meta = shift;
 343
 344         # e.g. titlesort (also makes sense for author)
 345         my $key = $pagestate{$page}{meta}{$meta . "sort"};
 346         return $key if defined $key;
 347
 348         # e.g. title
 349         $key = $pagestate{$page}{meta}{$meta};
 350         return $key if defined $key;
 351
 352         # fall back to closer-to-core things
 353         if ($meta eq 'title') {
 354                 return pagetitle(IkiWiki::basename($page));
 355         }
 356         elsif ($meta eq 'date') {
 357                 return $IkiWiki::pagectime{$page};
 358         }
 359         elsif ($meta eq 'updated') {
 360                 return $IkiWiki::pagemtime{$page};
 361         }
 362         else {
 363                 return '';
 364         }
 365 }
 366
 367 sub match {
 368         my $field=shift;
 369         my $page=shift;
 370
 371         # turn glob into a safe regexp
 372         my $re=IkiWiki::glob2re(shift);
 373
 374         my $val;
 375         if (exists $pagestate{$page}{meta}{$field}) {
 376                 $val=$pagestate{$page}{meta}{$field};
 377         }
 378         elsif ($field eq 'title') {
 379                 $val = pagetitle($page);
 380         }
 381
 382         if (defined $val) {
 383                 if ($val=~$re) {
 384                         return IkiWiki::SuccessReason->new("$re matches $field of $page", $page => $IkiWiki::DEPEND_CONTENT, "" => 1);
 385                 }
 386                 else {
 387                         return IkiWiki::FailReason->new("$re does not match $field of $page", $page => $IkiWiki::DEPEND_CONTENT, "" => 1);
 388                 }
 389         }
 390         else {
 391                 return IkiWiki::FailReason->new("$page does not have a $field", $page => $IkiWiki::DEPEND_CONTENT);
 392         }
 393 }
 394
 395 package IkiWiki::PageSpec;
 396
 397 sub match_title ($$;@) {
 398         IkiWiki::Plugin::meta::match("title", @_);
 399 }
 400
 401 sub match_author ($$;@) {
 402         IkiWiki::Plugin::meta::match("author", @_);
 403 }
 404
 405 sub match_authorurl ($$;@) {
 406         IkiWiki::Plugin::meta::match("authorurl", @_);
 407 }
 408
 409 sub match_license ($$;@) {
 410         IkiWiki::Plugin::meta::match("license", @_);
 411 }
 412
 413 sub match_copyright ($$;@) {
 414         IkiWiki::Plugin::meta::match("copyright", @_);
 415 }
 416
 417 sub match_guid ($$;@) {
 418         IkiWiki::Plugin::meta::match("guid", @_);
 419 }
 420
 421 package IkiWiki::SortSpec;
 422
 423 sub cmp_meta {
 424         my $meta = shift;
 425         error(gettext("sort=meta requires a parameter")) unless defined $meta;
 426
 427         if ($meta eq 'updated' || $meta eq 'date') {
 428                 return IkiWiki::Plugin::meta::get_sort_key($a, $meta)
 429                         <=>
 430                         IkiWiki::Plugin::meta::get_sort_key($b, $meta);
 431         }
 432
 433         return IkiWiki::Plugin::meta::get_sort_key($a, $meta)
 434                 cmp
 435                 IkiWiki::Plugin::meta::get_sort_key($b, $meta);
 436 }
 437
 438 # A prototype of how sort=title could behave in 4.0 or something
 439 sub cmp_meta_title {
 440         $_[0] = 'title';
 441         return cmp_meta(@_);
 442 }
 443
 444 1