feed.php

   1 <?php
   2 /**
   3  * XML feed export
   4  *
   5  * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
   6  * @author     Andreas Gohr <andi@splitbrain.org>
   7  *
   8  * @global array $conf
   9  * @global Input $INPUT
  10  */
  11
  12 if(!defined('DOKU_INC')) define('DOKU_INC', dirname(__FILE__).'/');
  13 require_once(DOKU_INC.'inc/init.php');
  14
  15 //close session
  16 session_write_close();
  17
  18 //feed disabled?
  19 if(!actionOK('rss')) {
  20     http_status(404);
  21     echo '<error>RSS feed is disabled.</error>';
  22     exit;
  23 }
  24
  25 // get params
  26 $opt = rss_parseOptions();
  27
  28 // the feed is dynamic - we need a cache for each combo
  29 // (but most people just use the default feed so it's still effective)
  30 $key   = join('', array_values($opt)).'$'.$_SERVER['REMOTE_USER'].'$'.$_SERVER['HTTP_HOST'].$_SERVER['SERVER_PORT'];
  31 $cache = new cache($key, '.feed');
  32
  33 // prepare cache depends
  34 $depends['files'] = getConfigFiles('main');
  35 $depends['age']   = $conf['rss_update'];
  36 $depends['purge'] = $INPUT->bool('purge');
  37
  38 // check cacheage and deliver if nothing has changed since last
  39 // time or the update interval has not passed, also handles conditional requests
  40 header('Cache-Control: must-revalidate, post-check=0, pre-check=0');
  41 header('Pragma: public');
  42 header('Content-Type: application/xml; charset=utf-8');
  43 header('X-Robots-Tag: noindex');
  44 if($cache->useCache($depends)) {
  45     http_conditionalRequest($cache->_time);
  46     if($conf['allowdebug']) header("X-CacheUsed: $cache->cache");
  47     print $cache->retrieveCache();
  48     exit;
  49 } else {
  50     http_conditionalRequest(time());
  51 }
  52
  53 // create new feed
  54 $rss                 = new DokuWikiFeedCreator();
  55 $rss->title          = $conf['title'].(($opt['namespace']) ? ' '.$opt['namespace'] : '');
  56 $rss->link           = DOKU_URL;
  57 $rss->syndicationURL = DOKU_URL.'feed.php';
  58 $rss->cssStyleSheet  = DOKU_URL.'lib/exe/css.php?s=feed';
  59
  60 $image        = new FeedImage();
  61 $image->title = $conf['title'];
  62 $image->url   = tpl_getMediaFile(array(':wiki:favicon.ico', ':favicon.ico', 'images/favicon.ico'), true);
  63 $image->link  = DOKU_URL;
  64 $rss->image   = $image;
  65
  66 $data  = null;
  67 $modes = array(
  68     'list'   => 'rssListNamespace',
  69     'search' => 'rssSearch',
  70     'recent' => 'rssRecentChanges'
  71 );
  72 if(isset($modes[$opt['feed_mode']])) {
  73     $data = $modes[$opt['feed_mode']]($opt);
  74 } else {
  75     $eventData = array(
  76         'opt'  => &$opt,
  77         'data' => &$data,
  78     );
  79     $event     = new Doku_Event('FEED_MODE_UNKNOWN', $eventData);
  80     if($event->advise_before(true)) {
  81         echo sprintf('<error>Unknown feed mode %s</error>', hsc($opt['feed_mode']));
  82         exit;
  83     }
  84     $event->advise_after();
  85 }
  86
  87 rss_buildItems($rss, $data, $opt);
  88 $feed = $rss->createFeed($opt['feed_type'], 'utf-8');
  89
  90 // save cachefile
  91 $cache->storeCache($feed);
  92
  93 // finally deliver
  94 print $feed;
  95
  96 // ---------------------------------------------------------------- //
  97
  98 /**
  99  * Get URL parameters and config options and return an initialized option array
 100  *
 101  * @author Andreas Gohr <andi@splitbrain.org>
 102  */
 103 function rss_parseOptions() {
 104     global $conf;
 105     global $INPUT;
 106
 107     $opt = array();
 108
 109     foreach(array(
 110                 // Basic feed properties
 111                 // Plugins may probably want to add new values to these
 112                 // properties for implementing own feeds
 113
 114                 // One of: list, search, recent
 115                 'feed_mode'    => array('str', 'mode', 'recent'),
 116                 // One of: diff, page, rev, current
 117                 'link_to'      => array('str', 'linkto', $conf['rss_linkto']),
 118                 // One of: abstract, diff, htmldiff, html
 119                 'item_content' => array('str', 'content', $conf['rss_content']),
 120
 121                 // Special feed properties
 122                 // These are only used by certain feed_modes
 123
 124                 // String, used for feed title, in list and rc mode
 125                 'namespace'    => array('str', 'ns', null),
 126                 // Positive integer, only used in rc mode
 127                 'items'        => array('int', 'num', $conf['recent']),
 128                 // Boolean, only used in rc mode
 129                 'show_minor'   => array('bool', 'minor', false),
 130                 // String, only used in search mode
 131                 'search_query' => array('str', 'q', null),
 132                 // One of: pages, media, both
 133                 'content_type' => array('str', 'view', $conf['rss_media'])
 134
 135             ) as $name => $val) {
 136         $opt[$name] = $INPUT->$val[0]($val[1], $val[2], true);
 137     }
 138
 139     $opt['items']      = max(0, (int) $opt['items']);
 140     $opt['show_minor'] = (bool) $opt['show_minor'];
 141
 142     $opt['guardmail'] = ($conf['mailguard'] != '' && $conf['mailguard'] != 'none');
 143
 144     $type = $INPUT->valid(
 145         'type',
 146         array( 'rss', 'rss2', 'atom', 'atom1', 'rss1'),
 147         $conf['rss_type']
 148     );
 149     switch($type) {
 150         case 'rss':
 151             $opt['feed_type'] = 'RSS0.91';
 152             $opt['mime_type'] = 'text/xml';
 153             break;
 154         case 'rss2':
 155             $opt['feed_type'] = 'RSS2.0';
 156             $opt['mime_type'] = 'text/xml';
 157             break;
 158         case 'atom':
 159             $opt['feed_type'] = 'ATOM0.3';
 160             $opt['mime_type'] = 'application/xml';
 161             break;
 162         case 'atom1':
 163             $opt['feed_type'] = 'ATOM1.0';
 164             $opt['mime_type'] = 'application/atom+xml';
 165             break;
 166         default:
 167             $opt['feed_type'] = 'RSS1.0';
 168             $opt['mime_type'] = 'application/xml';
 169     }
 170
 171     $eventData = array(
 172         'opt' => &$opt,
 173     );
 174     trigger_event('FEED_OPTS_POSTPROCESS', $eventData);
 175     return $opt;
 176 }
 177
 178 /**
 179  * Add recent changed pages to a feed object
 180  *
 181  * @author Andreas Gohr <andi@splitbrain.org>
 182  * @param  FeedCreator $rss the FeedCreator Object
 183  * @param  array       $data the items to add
 184  * @param  array       $opt  the feed options
 185  */
 186 function rss_buildItems(&$rss, &$data, $opt) {
 187     global $conf;
 188     global $lang;
 189     /* @var DokuWiki_Auth_Plugin $auth */
 190     global $auth;
 191
 192     $eventData = array(
 193         'rss'  => &$rss,
 194         'data' => &$data,
 195         'opt'  => &$opt,
 196     );
 197     $event     = new Doku_Event('FEED_DATA_PROCESS', $eventData);
 198     if($event->advise_before(false)) {
 199         foreach($data as $ditem) {
 200             if(!is_array($ditem)) {
 201                 // not an array? then only a list of IDs was given
 202                 $ditem = array('id' => $ditem);
 203             }
 204
 205             $item = new FeedItem();
 206             $id   = $ditem['id'];
 207             if(!$ditem['media']) {
 208                 $meta = p_get_metadata($id);
 209             } else {
 210                 $meta = array();
 211             }
 212
 213             // add date
 214             if($ditem['date']) {
 215                 $date = $ditem['date'];
 216             } elseif ($ditem['media']) {
 217                 $date = @filemtime(mediaFN($id));
 218             } elseif (@file_exists(wikiFN($id))) {
 219                 $date = @filemtime(wikiFN($id));
 220             } elseif($meta['date']['modified']) {
 221                 $date = $meta['date']['modified'];
 222             } else {
 223                 $date = 0;
 224             }
 225             if($date) $item->date = date('r', $date);
 226
 227             // add title
 228             if($conf['useheading'] && $meta['title']) {
 229                 $item->title = $meta['title'];
 230             } else {
 231                 $item->title = $ditem['id'];
 232             }
 233             if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
 234                 $item->title .= ' - '.strip_tags($ditem['sum']);
 235             }
 236
 237             // add item link
 238             switch($opt['link_to']) {
 239                 case 'page':
 240                     if($ditem['media']) {
 241                         $item->link = media_managerURL(
 242                             array(
 243                                  'image' => $id,
 244                                  'ns'    => getNS($id),
 245                                  'rev'   => $date
 246                             ), '&', true
 247                         );
 248                     } else {
 249                         $item->link = wl($id, 'rev='.$date, true, '&');
 250                     }
 251                     break;
 252                 case 'rev':
 253                     if($ditem['media']) {
 254                         $item->link = media_managerURL(
 255                             array(
 256                                  'image'       => $id,
 257                                  'ns'          => getNS($id),
 258                                  'rev'         => $date,
 259                                  'tab_details' => 'history'
 260                             ), '&', true
 261                         );
 262                     } else {
 263                         $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
 264                     }
 265                     break;
 266                 case 'current':
 267                     if($ditem['media']) {
 268                         $item->link = media_managerURL(
 269                             array(
 270                                  'image' => $id,
 271                                  'ns'    => getNS($id)
 272                             ), '&', true
 273                         );
 274                     } else {
 275                         $item->link = wl($id, '', true, '&');
 276                     }
 277                     break;
 278                 case 'diff':
 279                 default:
 280                     if($ditem['media']) {
 281                         $item->link = media_managerURL(
 282                             array(
 283                                  'image'       => $id,
 284                                  'ns'          => getNS($id),
 285                                  'rev'         => $date,
 286                                  'tab_details' => 'history',
 287                                  'mediado'     => 'diff'
 288                             ), '&', true
 289                         );
 290                     } else {
 291                         $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
 292                     }
 293             }
 294
 295             // add item content
 296             switch($opt['item_content']) {
 297                 case 'diff':
 298                 case 'htmldiff':
 299                     if($ditem['media']) {
 300                         $medialog = new MediaChangeLog($id);
 301                         $revs  = $medialog->getRevisions(0, 1);
 302                         $rev   = $revs[0];
 303                         $src_r = '';
 304                         $src_l = '';
 305
 306                         if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)), 300)) {
 307                             $more  = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
 308                             $src_r = ml($id, $more, true, '&amp;', true);
 309                         }
 310                         if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
 311                             $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
 312                             $src_l = ml($id, $more, true, '&amp;', true);
 313                         }
 314                         $content = '';
 315                         if($src_r) {
 316                             $content = '<table>';
 317                             $content .= '<tr><th width="50%">'.$rev.'</th>';
 318                             $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
 319                             $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
 320                             $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
 321                             $content .= '</table>';
 322                         }
 323
 324                     } else {
 325                         require_once(DOKU_INC.'inc/DifferenceEngine.php');
 326                         $pagelog = new PageChangeLog($id);
 327                         $revs = $pagelog->getRevisions(0, 1);
 328                         $rev  = $revs[0];
 329
 330                         if($rev) {
 331                             $df = new Diff(explode("\n", rawWiki($id, $rev)),
 332                                            explode("\n", rawWiki($id, '')));
 333                         } else {
 334                             $df = new Diff(array(''),
 335                                            explode("\n", rawWiki($id, '')));
 336                         }
 337
 338                         if($opt['item_content'] == 'htmldiff') {
 339                             // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
 340                             $tdf     = new TableDiffFormatter();
 341                             $content = '<table>';
 342                             $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
 343                             $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
 344                             $content .= $tdf->format($df);
 345                             $content .= '</table>';
 346                         } else {
 347                             // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
 348                             $udf     = new UnifiedDiffFormatter();
 349                             $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
 350                         }
 351                     }
 352                     break;
 353                 case 'html':
 354                     if($ditem['media']) {
 355                         if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
 356                             $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
 357                             $src     = ml($id, $more, true, '&amp;', true);
 358                             $content = '<img src="'.$src.'" alt="'.$id.'" />';
 359                         } else {
 360                             $content = '';
 361                         }
 362                     } else {
 363                         if (@filemtime(wikiFN($id)) === $date) {
 364                             $content = p_wiki_xhtml($id, '', false);
 365                         } else {
 366                             $content = p_wiki_xhtml($id, $date, false);
 367                         }
 368                         // no TOC in feeds
 369                         $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
 370
 371                         // add alignment for images
 372                         $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
 373                         $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
 374
 375                         // make URLs work when canonical is not set, regexp instead of rerendering!
 376                         if(!$conf['canonical']) {
 377                             $base    = preg_quote(DOKU_REL, '/');
 378                             $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
 379                         }
 380                     }
 381
 382                     break;
 383                 case 'abstract':
 384                 default:
 385                     if($ditem['media']) {
 386                         if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
 387                             $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
 388                             $src     = ml($id, $more, true, '&amp;', true);
 389                             $content = '<img src="'.$src.'" alt="'.$id.'" />';
 390                         } else {
 391                             $content = '';
 392                         }
 393                     } else {
 394                         $content = $meta['description']['abstract'];
 395                     }
 396             }
 397             $item->description = $content; //FIXME a plugin hook here could be senseful
 398
 399             // add user
 400             # FIXME should the user be pulled from metadata as well?
 401             $user         = @$ditem['user']; // the @ spares time repeating lookup
 402             $item->author = '';
 403             if($user && $conf['useacl'] && $auth) {
 404                 $userInfo = $auth->getUserData($user);
 405                 if($userInfo) {
 406                     switch($conf['showuseras']) {
 407                         case 'username':
 408                             $item->author = $userInfo['name'];
 409                             break;
 410                         default:
 411                             $item->author = $user;
 412                             break;
 413                     }
 414                 } else {
 415                     $item->author = $user;
 416                 }
 417                 if($userInfo && !$opt['guardmail']) {
 418                     $item->authorEmail = $userInfo['mail'];
 419                 } else {
 420                     //cannot obfuscate because some RSS readers may check validity
 421                     $item->authorEmail = $user.'@'.$ditem['ip'];
 422                 }
 423             } elseif($user) {
 424                 // this happens when no ACL but some Apache auth is used
 425                 $item->author      = $user;
 426                 $item->authorEmail = $user.'@'.$ditem['ip'];
 427             } else {
 428                 $item->authorEmail = 'anonymous@'.$ditem['ip'];
 429             }
 430
 431             // add category
 432             if(isset($meta['subject'])) {
 433                 $item->category = $meta['subject'];
 434             } else {
 435                 $cat = getNS($id);
 436                 if($cat) $item->category = $cat;
 437             }
 438
 439             // finally add the item to the feed object, after handing it to registered plugins
 440             $evdata = array(
 441                 'item'  => &$item,
 442                 'opt'   => &$opt,
 443                 'ditem' => &$ditem,
 444                 'rss'   => &$rss
 445             );
 446             $evt    = new Doku_Event('FEED_ITEM_ADD', $evdata);
 447             if($evt->advise_before()) {
 448                 $rss->addItem($item);
 449             }
 450             $evt->advise_after(); // for completeness
 451         }
 452     }
 453     $event->advise_after();
 454 }
 455
 456 /**
 457  * Add recent changed pages to the feed object
 458  *
 459  * @author Andreas Gohr <andi@splitbrain.org>
 460  */
 461 function rssRecentChanges($opt) {
 462     global $conf;
 463     $flags = RECENTS_SKIP_DELETED;
 464     if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
 465     if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
 466     if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
 467
 468     $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
 469     return $recents;
 470 }
 471
 472 /**
 473  * Add all pages of a namespace to the feed object
 474  *
 475  * @author Andreas Gohr <andi@splitbrain.org>
 476  */
 477 function rssListNamespace($opt) {
 478     require_once(DOKU_INC.'inc/search.php');
 479     global $conf;
 480
 481     $ns = ':'.cleanID($opt['namespace']);
 482     $ns = str_replace(':', '/', $ns);
 483
 484     $data = array();
 485     $search_opts = array(
 486         'depth' => 1,
 487         'pagesonly' => true,
 488         'listfiles' => true
 489     );
 490     search($data, $conf['datadir'], 'search_universal', $search_opts, $ns);
 491
 492     return $data;
 493 }
 494
 495 /**
 496  * Add the result of a full text search to the feed object
 497  *
 498  * @author Andreas Gohr <andi@splitbrain.org>
 499  */
 500 function rssSearch($opt) {
 501     if(!$opt['search_query']) return array();
 502
 503     require_once(DOKU_INC.'inc/fulltext.php');
 504     $data = ft_pageSearch($opt['search_query'], $poswords);
 505     $data = array_keys($data);
 506
 507     return $data;
 508 }
 509
 510 //Setup VIM: ex: et ts=4 :