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                 push @{$links{$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                 if (exists $params{server} && safeurl($params{server})) {
 167                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
 168                                 '" rel="openid.server" />';
 169                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($params{server}).
 170                                 '" rel="openid2.provider" />';
 171                 }
 172                 if (safeurl($value)) {
 173                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
 174                                 '" rel="openid.delegate" />';
 175                         push @{$metaheaders{$page}}, '<link href="'.encode_entities($value).
 176                                 '" rel="openid2.local_id" />';
 177                 }
 178                 if (exists $params{"xrds-location"} && safeurl($params{"xrds-location"})) {
 179                         push @{$metaheaders{$page}}, '<meta http-equiv="X-XRDS-Location"'.
 180                                 'content="'.encode_entities($params{"xrds-location"}).'" />';
 181                 }
 182         }
 183         elsif ($key eq 'redir') {
 184                 return "" if $page ne $destpage;
 185                 my $safe=0;
 186                 if ($value !~ /^\w+:\/\//) {
 187                         my ($redir_page, $redir_anchor) = split /\#/, $value;
 188
 189                         add_depends($page, $redir_page);
 190                         my $link=bestlink($page, $redir_page);
 191                         if (! length $link) {
 192                                 error gettext("redir page not found")
 193                         }
 194
 195                         $value=urlto($link, $page);
 196                         $value.='#'.$redir_anchor if defined $redir_anchor;
 197                         $safe=1;
 198
 199                         # redir cycle detection
 200                         $pagestate{$page}{meta}{redir}=$link;
 201                         my $at=$page;
 202                         my %seen;
 203                         while (exists $pagestate{$at}{meta}{redir}) {
 204                                 if ($seen{$at}) {
 205                                         error gettext("redir cycle is not allowed")
 206                                 }
 207                                 $seen{$at}=1;
 208                                 $at=$pagestate{$at}{meta}{redir};
 209                         }
 210                 }
 211                 else {
 212                         $value=encode_entities($value);
 213                 }
 214                 my $delay=int(exists $params{delay} ? $params{delay} : 0);
 215                 my $redir="<meta http-equiv=\"refresh\" content=\"$delay; URL=$value\" />";
 216                 if (! $safe) {
 217                         $redir=scrub($redir, $destpage);
 218                 }
 219                 push @{$metaheaders{$page}}, $redir;
 220         }
 221         elsif ($key eq 'link') {
 222                 if (%params) {
 223                         push @{$metaheaders{$page}}, scrub("<link href=\"".encode_entities($value)."\" ".
 224                                 join(" ", map {
 225                                         encode_entities($_)."=\"".encode_entities(decode_entities($params{$_}))."\""
 226                                 } keys %params).
 227                                 " />\n", $destpage);
 228                 }
 229         }
 230         elsif ($key eq 'robots') {
 231                 push @{$metaheaders{$page}}, '<meta name="robots"'.
 232                         ' content="'.encode_entities($value).'" />';
 233         }
 234         else {
 235                 push @{$metaheaders{$page}}, scrub('<meta name="'.encode_entities($key).
 236                         '" content="'.encode_entities($value).'" />', $destpage);
 237         }
 238
 239         return "";
 240 }
 241
 242 sub pagetemplate (@) {
 243         my %params=@_;
 244         my $page=$params{page};
 245         my $destpage=$params{destpage};
 246         my $template=$params{template};
 247
 248         if (exists $metaheaders{$page} && $template->query(name => "meta")) {
 249                 # avoid duplicate meta lines
 250                 my %seen;
 251                 $template->param(meta => join("\n", grep { (! $seen{$_}) && ($seen{$_}=1) } @{$metaheaders{$page}}));
 252         }
 253         if (exists $pagestate{$page}{meta}{title} && $template->query(name => "title")) {
 254                 $template->param(title => $pagestate{$page}{meta}{title});
 255                 $template->param(title_overridden => 1);
 256         }
 257
 258         foreach my $field (qw{author authorurl permalink}) {
 259                 $template->param($field => $pagestate{$page}{meta}{$field})
 260                         if exists $pagestate{$page}{meta}{$field} && $template->query(name => $field);
 261         }
 262
 263         foreach my $field (qw{license copyright}) {
 264                 if (exists $pagestate{$page}{meta}{$field} && $template->query(name => $field) &&
 265                     ($page eq $destpage || ! exists $pagestate{$destpage}{meta}{$field} ||
 266                      $pagestate{$page}{meta}{$field} ne $pagestate{$destpage}{meta}{$field})) {
 267                         $template->param($field => htmlize($page, $destpage, $pagestate{$page}{meta}{$field}));
 268                 }
 269         }
 270 }
 271
 272 sub match {
 273         my $field=shift;
 274         my $page=shift;
 275
 276         # turn glob into a safe regexp
 277         my $re=IkiWiki::glob2re(shift);
 278
 279         my $val;
 280         if (exists $pagestate{$page}{meta}{$field}) {
 281                 $val=$pagestate{$page}{meta}{$field};
 282         }
 283         elsif ($field eq 'title') {
 284                 $val = pagetitle($page);
 285         }
 286
 287         if (defined $val) {
 288                 if ($val=~/^$re$/i) {
 289                         return IkiWiki::SuccessReason->new("$re matches $field of $page");
 290                 }
 291                 else {
 292                         return IkiWiki::FailReason->new("$re does not match $field of $page");
 293                 }
 294         }
 295         else {
 296                 return IkiWiki::FailReason->new("$page does not have a $field");
 297         }
 298 }
 299
 300 package IkiWiki::PageSpec;
 301
 302 sub match_title ($$;@) {
 303         IkiWiki::Plugin::meta::match("title", @_);
 304 }
 305
 306 sub match_author ($$;@) {
 307         IkiWiki::Plugin::meta::match("author", @_);
 308 }
 309
 310 sub match_authorurl ($$;@) {
 311         IkiWiki::Plugin::meta::match("authorurl", @_);
 312 }
 313
 314 sub match_license ($$;@) {
 315         IkiWiki::Plugin::meta::match("license", @_);
 316 }
 317
 318 sub match_copyright ($$;@) {
 319         IkiWiki::Plugin::meta::match("copyright", @_);
 320 }
 321
 322 1