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                 },
  24 }
  25
  26 sub needsbuild (@) {
  27         my $needsbuild=shift;
  28         foreach my $page (keys %pagestate) {
  29                 if (exists $pagestate{$page}{meta}) {
  30                         if (exists $pagesources{$page} &&
  31                             grep { $_ eq $pagesources{$page} } @$needsbuild) {
  32                                 # remove state, it will be re-added
  33                                 # if the preprocessor directive is still
  34                                 # there during the rebuild
  35                                 delete $pagestate{$page}{meta};
  36                         }
  37                 }
  38         }
  39 }
  40
  41 sub scrub ($$) {
  42         if (IkiWiki::Plugin::htmlscrubber->can("sanitize")) {
  43                 return IkiWiki::Plugin::htmlscrubber::sanitize(
  44                         content => shift, destpage => shift);
  45         }
  46         else {
  47                 return shift;
  48         }
  49 }
  50
  51 sub safeurl ($) {
  52         my $url=shift;
  53         if (exists $IkiWiki::Plugin::htmlscrubber::{safe_url_regexp} &&
  54             defined $IkiWiki::Plugin::htmlscrubber::safe_url_regexp) {
  55                 return $url=~/$IkiWiki::Plugin::htmlscrubber::safe_url_regexp/;
  56         }
  57         else {
  58                 return 1;
  59         }
  60 }
  61
  62 sub htmlize ($$$) {
  63         my $page = shift;
  64         my $destpage = shift;
  65
  66         return IkiWiki::htmlize($page, $destpage, pagetype($pagesources{$page}),
  67                 IkiWiki::linkify($page, $destpage,
  68                 IkiWiki::preprocess($page, $destpage, shift)));
  69 }
  70
  71 sub preprocess (@) {
  72         return "" unless @_;
  73         my %params=@_;
  74         my $key=shift;
  75         my $value=$params{$key};
  76         delete $params{$key};
  77         my $page=$params{page};
  78         delete $params{page};
  79         my $destpage=$params{destpage};
  80         delete $params{destpage};
  81         delete $params{preview};
  82
  83         eval q{use HTML::Entities};
  84         # Always decode, even if encoding later, since it might not be
  85         # fully encoded.
  86         $value=decode_entities($value);
  87
  88         # Metadata collection that needs to happen during the scan pass.
  89         if ($key eq 'title') {
  90                 $pagestate{$page}{meta}{title}=HTML::Entities::encode_numeric($value);
  91                 # fallthrough
  92         }
  93         elsif ($key eq 'description') {
  94                 $pagestate{$page}{meta}{description}=HTML::Entities::encode_numeric($value);
  95                 # fallthrough
  96         }
  97         elsif ($key eq 'guid') {
  98                 $pagestate{$page}{meta}{guid}=HTML::Entities::encode_numeric($value);
  99                 # fallthrough
 100         }
 101         elsif ($key eq 'license') {
 102                 push @{$metaheaders{$page}}, '<link rel="license" href="#page_license" />';
 103                 $pagestate{$page}{meta}{license}=$value;
 104                 return "";
 105         }
 106         elsif ($key eq 'copyright') {
 107                 push @{$metaheaders{$page}}, '<link rel="copyright" href="#page_copyright" />';
 108                 $pagestate{$page}{meta}{copyright}=$value;
 109                 return "";
 110         }
 111         elsif ($key eq 'link' && ! %params) {
 112                 # hidden WikiLink
 113                 add_link($page, $value);
 114                 return "";
 115         }
 116         elsif ($key eq 'author') {
 117                 $pagestate{$page}{meta}{author}=$value;
 118                 # fallthorough
 119         }
 120         elsif ($key eq 'authorurl') {
 121                 $pagestate{$page}{meta}{authorurl}=$value if safeurl($value);
 122                 # fallthrough
 123         }
 124         elsif ($key eq 'date') {
 125                 eval q{use Date::Parse};
 126                 if (! $@) {
 127                         my $time = str2time($value);
 128                         $IkiWiki::pagectime{$page}=$time if defined $time;
 129                 }
 130         }
 131         elsif ($key eq 'updated') {
 132                 eval q{use Date::Parse};
 133                 if (! $@) {
 134                         my $time = str2time($value);
 135                         $pagestate{$page}{meta}{updated}=$time if defined $time;
 136                 }
 137         }
 138
 139         if (! defined wantarray) {
 140                 # avoid collecting duplicate data during scan pass
 141                 return;
 142         }
 143
 144         # Metadata collection that happens only during preprocessing pass.
 145         if ($key eq 'permalink') {
 146                 if (safeurl($value)) {
 147                         $pagestate{$page}{meta}{permalink}=$value;
 148                         push @{$metaheaders{$page}}, scrub('<link rel="bookmark" href="'.encode_entities($value).'" />', $destpage);
 149                 }
 150         }
 151         elsif ($key eq 'stylesheet') {
 152                 my $rel=exists $params{rel} ? $params{rel} : "alternate stylesheet";
 153                 my $title=exists $params{title} ? $params{title} : $value;
 154                 # adding .css to the value prevents using any old web
 155                 # editable page as a stylesheet
 156                 my $stylesheet=bestlink($page, $value.".css");
 157                 if (! length $stylesheet) {
 158                         error gettext("stylesheet not found")
 159                 }
 160                 push @{$metaheaders{$page}}, '<link href="'.urlto($stylesheet, $page).
 161                         '" rel="'.encode_entities($rel).
 162                         '" title="'.encode_entities($title).
 163                         "\" type=\"text/css\" />";
 164         }
 165         elsif ($key eq 'openid') {
 166                 my $delegate=0; # both by default
 167                 if (exists $params{delegate}) {
 168                         $delegate = 1 if lc $params{delegate} eq 'openid';
 169                         $delegate = 2 if lc $params{delegate} eq 'openid2';
 170                 }
 171                 if (exists $params{server} && safeurl($params{server})) {
 172                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
 173                                 '" rel="openid.server" />' if $delegate ne 2;
 174                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
 175                                 '" rel="openid2.provider" />' if $delegate ne 1;
 176                 }
 177                 if (safeurl($value)) {
 178                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
 179                                 '" rel="openid.delegate" />' if $delegate ne 2;
 180                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
 181                                 '" rel="openid2.local_id" />' if $delegate ne 1;
 182                 }
 183                 if (exists $params{"xrds-location"} && safeurl($params{"xrds-location"})) {
 184                         push @{$metaheaders{$page}}, '<meta http-equiv="X-XRDS-Location"'.
 185                                 'content="'.encode_entities($params{"xrds-location"}).'" />';
 186                 }
 187         }
 188         elsif ($key eq 'redir') {
 189                 return "" if $page ne $destpage;
 190                 my $safe=0;
 191                 if ($value !~ /^\w+:\/\//) {
 192                         my ($redir_page, $redir_anchor) = split /\#/, $value;
 193
 194                         my $link=bestlink($page, $redir_page);
 195                         if (! length $link) {
 196                                 error gettext("redir page not found")
 197                         }
 198                         add_depends($page, $link, presence => 1);
 199
 200                         $value=urlto($link, $page);
 201                         $value.='#'.$redir_anchor if defined $redir_anchor;
 202                         $safe=1;
 203
 204                         # redir cycle detection
 205                         $pagestate{$page}{meta}{redir}=$link;
 206                         my $at=$page;
 207                         my %seen;
 208                         while (exists $pagestate{$at}{meta}{redir}) {
 209                                 if ($seen{$at}) {
 210                                         error gettext("redir cycle is not allowed")
 211                                 }
 212                                 $seen{$at}=1;
 213                                 $at=$pagestate{$at}{meta}{redir};
 214                         }
 215                 }
 216                 else {
 217                         $value=encode_entities($value);
 218                 }
 219                 my $delay=int(exists $params{delay} ? $params{delay} : 0);
 220                 my $redir="<meta http-equiv=\"refresh\" content=\"$delay; URL=$value\" />";
 221                 if (! $safe) {
 222                         $redir=scrub($redir, $destpage);
 223                 }
 224                 push @{$metaheaders{$page}}, $redir;
 225         }
 226         elsif ($key eq 'link') {
 227                 if (%params) {
 228                         push @{$metaheaders{$page}}, scrub("<link href=\"".encode_entities($value)."\" ".
 229                                 join(" ", map {
 230                                         encode_entities($_)."=\"".encode_entities(decode_entities($params{$_}))."\""
 231                                 } keys %params).
 232                                 " />\n", $destpage);
 233                 }
 234         }
 235         elsif ($key eq 'robots') {
 236                 push @{$metaheaders{$page}}, '<meta name="robots"'.
 237                         ' content="'.encode_entities($value).'" />';
 238         }
 239         else {
 240                 push @{$metaheaders{$page}}, scrub('<meta name="'.encode_entities($key).
 241                         '" content="'.encode_entities($value).'" />', $destpage);
 242         }
 243
 244         return "";
 245 }
 246
 247 sub pagetemplate (@) {
 248         my %params=@_;
 249         my $page=$params{page};
 250         my $destpage=$params{destpage};
 251         my $template=$params{template};
 252
 253         if (exists $metaheaders{$page} && $template->query(name => "meta")) {
 254                 # avoid duplicate meta lines
 255                 my %seen;
 256                 $template->param(meta => join("\n", grep { (! $seen{$_}) && ($seen{$_}=1) } @{$metaheaders{$page}}));
 257         }
 258         if (exists $pagestate{$page}{meta}{title} && $template->query(name => "title")) {
 259                 $template->param(title => $pagestate{$page}{meta}{title});
 260                 $template->param(title_overridden => 1);
 261         }
 262
 263         foreach my $field (qw{author authorurl permalink}) {
 264                 $template->param($field => $pagestate{$page}{meta}{$field})
 265                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
 266         }
 267
 268         foreach my $field (qw{license copyright}) {
 269                 if (exists $pagestate{$page}{meta}{$field} && $template->query(name => $field) &&
 270                     ($page eq $destpage || ! exists $pagestate{$destpage}{meta}{$field} ||
 271                      $pagestate{$page}{meta}{$field} ne $pagestate{$destpage}{meta}{$field})) {
 272                         $template->param($field => htmlize($page, $destpage, $pagestate{$page}{meta}{$field}));
 273                 }
 274         }
 275 }
 276
 277 sub match {
 278         my $field=shift;
 279         my $page=shift;
 280
 281         # turn glob into a safe regexp
 282         my $re=IkiWiki::glob2re(shift);
 283
 284         my $val;
 285         if (exists $pagestate{$page}{meta}{$field}) {
 286                 $val=$pagestate{$page}{meta}{$field};
 287         }
 288         elsif ($field eq 'title') {
 289                 $val = pagetitle($page);
 290         }
 291
 292         if (defined $val) {
 293                 if ($val=~/^$re$/i) {
 294                         return IkiWiki::SuccessReason->new("$re matches $field of $page", $page => $IkiWiki::DEPEND_CONTENT);
 295                 }
 296                 else {
 297                         return IkiWiki::FailReason->new("$re does not match $field of $page", $page => $IkiWiki::DEPEND_CONTENT);
 298                 }
 299         }
 300         else {
 301                 return IkiWiki::FailReason->new("$page does not have a $field", $page => $IkiWiki::DEPEND_CONTENT);
 302         }
 303 }
 304
 305 package IkiWiki::PageSpec;
 306
 307 sub match_title ($$;@) {
 308         IkiWiki::Plugin::meta::match("title", @_);
 309 }
 310
 311 sub match_author ($$;@) {
 312         IkiWiki::Plugin::meta::match("author", @_);
 313 }
 314
 315 sub match_authorurl ($$;@) {
 316         IkiWiki::Plugin::meta::match("authorurl", @_);
 317 }
 318
 319 sub match_license ($$;@) {
 320         IkiWiki::Plugin::meta::match("license", @_);
 321 }
 322
 323 sub match_copyright ($$;@) {
 324         IkiWiki::Plugin::meta::match("copyright", @_);
 325 }
 326
 327 1