includes/Metadata.php

   1 <?php
   2 /**
   3  * Metadata.php -- provides DublinCore and CreativeCommons metadata
   4  * Copyright 2004, Evan Prodromou <evan@wikitravel.org>.
   5  *
   6  *  This program is free software; you can redistribute it and/or modify
   7  *  it under the terms of the GNU General Public License as published by
   8  *  the Free Software Foundation; either version 2 of the License, or
   9  *  (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  *  GNU General Public License for more details.
  15  *
  16  *  You should have received a copy of the GNU General Public License
  17  *  along with this program; if not, write to the Free Software
  18  *  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  19  *
  20  * @author Evan Prodromou <evan@wikitravel.org>
  21  */
  22
  23 /**
  24  * TODO: Perhaps make this file into a Metadata class, with static methods (declared
  25  * as private where indicated), to move these functions out of the global namespace?
  26  */
  27 define('RDF_TYPE_PREFS', "application/rdf+xml,text/xml;q=0.7,application/xml;q=0.5,text/rdf;q=0.1");
  28
  29 function wfDublinCoreRdf($article) {
  30
  31         $url = dcReallyFullUrl($article->mTitle);
  32
  33         if (rdfSetup()) {
  34                 dcPrologue($url);
  35                 dcBasics($article);
  36                 dcEpilogue();
  37         }
  38 }
  39
  40 function wfCreativeCommonsRdf($article) {
  41
  42         if (rdfSetup()) {
  43                 global $wgRightsUrl;
  44
  45                 $url = dcReallyFullUrl($article->mTitle);
  46
  47                 ccPrologue();
  48                 ccSubPrologue('Work', $url);
  49                 dcBasics($article);
  50                 if (isset($wgRightsUrl)) {
  51                         $url = htmlspecialchars( $wgRightsUrl );
  52                         print "    <cc:license rdf:resource=\"$url\" />\n";
  53                 }
  54
  55                 ccSubEpilogue('Work');
  56
  57                 if (isset($wgRightsUrl)) {
  58                         $terms = ccGetTerms($wgRightsUrl);
  59                         if ($terms) {
  60                                 ccSubPrologue('License', $wgRightsUrl);
  61                                 ccLicense($terms);
  62                                 ccSubEpilogue('License');
  63                         }
  64                 }
  65         }
  66
  67         ccEpilogue();
  68 }
  69
  70 /**
  71  * @private
  72  */
  73 function rdfSetup() {
  74         global $wgOut, $_SERVER;
  75
  76         $httpaccept = isset($_SERVER['HTTP_ACCEPT']) ? $_SERVER['HTTP_ACCEPT'] : null;
  77
  78         $rdftype = wfNegotiateType(wfAcceptToPrefs($httpaccept), wfAcceptToPrefs(RDF_TYPE_PREFS));
  79
  80         if (!$rdftype) {
  81                 wfHttpError(406, "Not Acceptable", wfMsg("notacceptable"));
  82                 return false;
  83         } else {
  84                 $wgOut->disable();
  85                 header( "Content-type: {$rdftype}; charset=utf-8" );
  86                 $wgOut->sendCacheControl();
  87                 return true;
  88         }
  89 }
  90
  91 /**
  92  * @private
  93  */
  94 function dcPrologue($url) {
  95         global $wgOutputEncoding;
  96
  97         $url = htmlspecialchars( $url );
  98         print "<" . "?xml version=\"1.0\" encoding=\"{$wgOutputEncoding}\" ?" . ">
  99
 100                                                                                                                                                           <!DOCTYPE rdf:RDF PUBLIC \"-//DUBLIN CORE//DCMES DTD 2002/07/31//EN\" \"http://dublincore.org/documents/2002/07/31/dcmes-xml/dcmes-xml-dtd.dtd\">
 101
 102                                                                                                                                                           <rdf:RDF xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\"
 103                                                                                                                                                           xmlns:dc=\"http://purl.org/dc/elements/1.1/\">
 104                                                                                                                                                           <rdf:Description rdf:about=\"$url\">
 105                                                                                                                                                           ";
 106 }
 107
 108 /**
 109  * @private
 110  */
 111 function dcEpilogue() {
 112         print "
 113                         </rdf:Description>
 114                         </rdf:RDF>
 115                         ";
 116 }
 117
 118 /**
 119  * @private
 120  */
 121 function dcBasics($article) {
 122         global $wgContLanguageCode, $wgSitename;
 123
 124         dcElement('title', $article->mTitle->getText());
 125         dcPageOrString('publisher', wfMsg('aboutpage'), $wgSitename);
 126         dcElement('language', $wgContLanguageCode);
 127         dcElement('type', 'Text');
 128         dcElement('format', 'text/html');
 129         dcElement('identifier', dcReallyFullUrl($article->mTitle));
 130         dcElement('date', dcDate($article->getTimestamp()));
 131
 132         $last_editor = $article->getUser();
 133
 134         if ($last_editor == 0) {
 135                 dcPerson('creator', 0);
 136         } else {
 137                 dcPerson('creator', $last_editor, $article->getUserText(),
 138                                  User::whoIsReal($last_editor));
 139         }
 140
 141         $contributors = $article->getContributors();
 142
 143         foreach ($contributors as $user_parts) {
 144                 dcPerson('contributor', $user_parts[0], $user_parts[1], $user_parts[2]);
 145         }
 146
 147         dcRights();
 148 }
 149
 150 /**
 151  * @private
 152  */
 153 function ccPrologue() {
 154         global $wgOutputEncoding;
 155
 156         echo "<" . "?xml version='1.0'  encoding='{$wgOutputEncoding}' ?" . ">
 157
 158                                                                                                                                                   <rdf:RDF xmlns:cc=\"http://web.resource.org/cc/\"
 159                                                                                                                                                   xmlns:dc=\"http://purl.org/dc/elements/1.1/\"
 160                                                                                                                                                   xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\">
 161                                                                                                                                                   ";
 162 }
 163
 164 /**
 165  * @private
 166  */
 167 function ccSubPrologue($type, $url) {
 168         $url = htmlspecialchars( $url );
 169         echo "  <cc:{$type} rdf:about=\"{$url}\">\n";
 170 }
 171
 172 /**
 173  * @private
 174  */
 175 function ccSubEpilogue($type) {
 176         echo "  </cc:{$type}>\n";
 177 }
 178
 179 /**
 180  * @private
 181  */
 182 function ccLicense($terms) {
 183
 184         foreach ($terms as $term) {
 185                 switch ($term) {
 186                  case 're':
 187                         ccTerm('permits', 'Reproduction'); break;
 188                  case 'di':
 189                         ccTerm('permits', 'Distribution'); break;
 190                  case 'de':
 191                         ccTerm('permits', 'DerivativeWorks'); break;
 192                  case 'nc':
 193                         ccTerm('prohibits', 'CommercialUse'); break;
 194                  case 'no':
 195                         ccTerm('requires', 'Notice'); break;
 196                  case 'by':
 197                         ccTerm('requires', 'Attribution'); break;
 198                  case 'sa':
 199                         ccTerm('requires', 'ShareAlike'); break;
 200                  case 'sc':
 201                         ccTerm('requires', 'SourceCode'); break;
 202                 }
 203         }
 204 }
 205
 206 /**
 207  * @private
 208  */
 209 function ccTerm($term, $name) {
 210         print "    <cc:{$term} rdf:resource=\"http://web.resource.org/cc/{$name}\" />\n";
 211 }
 212
 213 /**
 214  * @private
 215  */
 216 function ccEpilogue() {
 217         echo "</rdf:RDF>\n";
 218 }
 219
 220 /**
 221  * @private
 222  */
 223 function dcElement($name, $value) {
 224         $value = htmlspecialchars( $value );
 225         print "    <dc:{$name}>{$value}</dc:{$name}>\n";
 226 }
 227
 228 /**
 229  * @private
 230  */
 231 function dcDate($timestamp) {
 232         return substr($timestamp, 0, 4) . '-'
 233           . substr($timestamp, 4, 2) . '-'
 234           . substr($timestamp, 6, 2);
 235 }
 236
 237 /**
 238  * @private
 239  */
 240 function dcReallyFullUrl($title) {
 241         return $title->getFullURL();
 242 }
 243
 244 /**
 245  * @private
 246  */
 247 function dcPageOrString($name, $page, $str) {
 248         $nt = Title::newFromText($page);
 249
 250         if (!$nt || $nt->getArticleID() == 0) {
 251                 dcElement($name, $str);
 252         } else {
 253                 dcPage($name, $nt);
 254         }
 255 }
 256
 257 /**
 258  * @private
 259  */
 260 function dcPage($name, $title) {
 261         dcUrl($name, dcReallyFullUrl($title));
 262 }
 263
 264 /**
 265  * @private
 266  */
 267 function dcUrl($name, $url) {
 268         $url = htmlspecialchars( $url );
 269         print "    <dc:{$name} rdf:resource=\"{$url}\" />\n";
 270 }
 271
 272 /**
 273  * @private
 274  */
 275 function dcPerson($name, $id, $user_name='', $user_real_name='') {
 276         global $wgContLang;
 277
 278         if ($id == 0) {
 279                 dcElement($name, wfMsg('anonymous'));
 280         } else if ( !empty($user_real_name) ) {
 281                 dcElement($name, $user_real_name);
 282         } else {
 283                 # XXX: This shouldn't happen.
 284                 if( empty( $user_name ) ) {
 285                         $user_name = User::whoIs($id);
 286                 }
 287                 dcPageOrString($name, $wgContLang->getNsText(NS_USER) . ':' . $user_name, wfMsg('siteuser', $user_name));
 288         }
 289 }
 290
 291 /**
 292  * Takes an arg, for future enhancement with different rights for
 293  * different pages.
 294  * @private
 295  */
 296 function dcRights() {
 297
 298         global $wgRightsPage, $wgRightsUrl, $wgRightsText;
 299
 300         if (isset($wgRightsPage) &&
 301                 ($nt = Title::newFromText($wgRightsPage))
 302                 && ($nt->getArticleID() != 0)) {
 303                 dcPage('rights', $nt);
 304         } else if (isset($wgRightsUrl)) {
 305                 dcUrl('rights', $wgRightsUrl);
 306         } else if (isset($wgRightsText)) {
 307                 dcElement('rights', $wgRightsText);
 308         }
 309 }
 310
 311 /**
 312  * @private
 313  */
 314 function ccGetTerms($url) {
 315         global $wgLicenseTerms;
 316
 317         if (isset($wgLicenseTerms)) {
 318                 return $wgLicenseTerms;
 319         } else {
 320                 $known = getKnownLicenses();
 321                 if( isset( $known[$url] ) ) {
 322                         return $known[$url];
 323                 } else {
 324                         return array();
 325                 }
 326         }
 327 }
 328
 329 /**
 330  * @private
 331  */
 332 function getKnownLicenses() {
 333
 334         $ccLicenses = array('by', 'by-nd', 'by-nd-nc', 'by-nc',
 335                                                 'by-nc-sa', 'by-sa');
 336         $ccVersions = array('1.0', '2.0');
 337         $knownLicenses = array();
 338
 339         foreach ($ccVersions as $version) {
 340                 foreach ($ccLicenses as $license) {
 341                         if( $version == '2.0' && substr( $license, 0, 2) != 'by' ) {
 342                                 # 2.0 dropped the non-attribs licenses
 343                                 continue;
 344                         }
 345                         $lurl = "http://creativecommons.org/licenses/{$license}/{$version}/";
 346                         $knownLicenses[$lurl] = explode('-', $license);
 347                         $knownLicenses[$lurl][] = 're';
 348                         $knownLicenses[$lurl][] = 'di';
 349                         $knownLicenses[$lurl][] = 'no';
 350                         if (!in_array('nd', $knownLicenses[$lurl])) {
 351                                 $knownLicenses[$lurl][] = 'de';
 352                         }
 353                 }
 354         }
 355
 356         /* Handle the GPL and LGPL, too. */
 357
 358         $knownLicenses['http://creativecommons.org/licenses/GPL/2.0/'] =
 359           array('de', 're', 'di', 'no', 'sa', 'sc');
 360         $knownLicenses['http://creativecommons.org/licenses/LGPL/2.1/'] =
 361           array('de', 're', 'di', 'no', 'sa', 'sc');
 362         $knownLicenses['http://www.gnu.org/copyleft/fdl.html'] =
 363           array('de', 're', 'di', 'no', 'sa', 'sc');
 364
 365         return $knownLicenses;
 366 }
 367
 368