xref: /dokuwiki/feed.php (revision a4bc205acc656841e5020ebf63fa1303ebb4f7d5)
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 $INPUT
10 */
11
12if(!defined('DOKU_INC')) define('DOKU_INC', dirname(__FILE__).'/');
13require_once(DOKU_INC.'inc/init.php');
14
15//close session
16session_write_close();
17
18//feed disabled?
19if(!actionOK('rss')) {
20    http_status(404);
21    echo '<error>RSS feed is disabled.</error>';
22    exit;
23}
24
25// get params
26$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'] = $INPUT->bool('purge');
37
38// check cacheage and deliver if nothing has changed since last
39// time or the update interval has not passed, also handles conditional requests
40header('Cache-Control: must-revalidate, post-check=0, pre-check=0');
41header('Pragma: public');
42header('Content-Type: application/xml; charset=utf-8');
43header('X-Robots-Tag: noindex');
44if($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$modes = array(
68    'list'   => 'rssListNamespace',
69    'search' => 'rssSearch',
70    'recent' => 'rssRecentChanges'
71);
72if(isset($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    $event->advise_after();
85}
86
87rss_buildItems($rss, $data, $opt);
88$feed = $rss->createFeed($opt['feed_type'], 'utf-8');
89
90// save cachefile
91$cache->storeCache($feed);
92
93// finally deliver
94print $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 */
103function 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 = valid_input_set(
145        'type', array(
146                     'rss', 'rss2', 'atom', 'atom1', 'rss1',
147                     'default' => $conf['rss_type']
148                ),
149        $_REQUEST
150    );
151    switch($type) {
152        case 'rss':
153            $opt['feed_type'] = 'RSS0.91';
154            $opt['mime_type'] = 'text/xml';
155            break;
156        case 'rss2':
157            $opt['feed_type'] = 'RSS2.0';
158            $opt['mime_type'] = 'text/xml';
159            break;
160        case 'atom':
161            $opt['feed_type'] = 'ATOM0.3';
162            $opt['mime_type'] = 'application/xml';
163            break;
164        case 'atom1':
165            $opt['feed_type'] = 'ATOM1.0';
166            $opt['mime_type'] = 'application/atom+xml';
167            break;
168        default:
169            $opt['feed_type'] = 'RSS1.0';
170            $opt['mime_type'] = 'application/xml';
171    }
172
173    $eventData = array(
174        'opt' => &$opt,
175    );
176    trigger_event('FEED_OPTS_POSTPROCESS', $eventData);
177    return $opt;
178}
179
180/**
181 * Add recent changed pages to a feed object
182 *
183 * @author Andreas Gohr <andi@splitbrain.org>
184 * @param  FeedCreator $rss the FeedCreator Object
185 * @param  array       $data the items to add
186 * @param  array       $opt  the feed options
187 */
188function rss_buildItems(&$rss, &$data, $opt) {
189    global $conf;
190    global $lang;
191    /* @var DokuWiki_Auth_Plugin $auth */
192    global $auth;
193
194    $eventData = array(
195        'rss'  => &$rss,
196        'data' => &$data,
197        'opt'  => &$opt,
198    );
199    $event     = new Doku_Event('FEED_DATA_PROCESS', $eventData);
200    if($event->advise_before(false)) {
201        foreach($data as $ditem) {
202            if(!is_array($ditem)) {
203                // not an array? then only a list of IDs was given
204                $ditem = array('id' => $ditem);
205            }
206
207            $item = new FeedItem();
208            $id   = $ditem['id'];
209            if(!$ditem['media']) {
210                $meta = p_get_metadata($id);
211            } else {
212                $meta = array();
213            }
214
215            // add date
216            if($ditem['date']) {
217                $date = $ditem['date'];
218            } elseif ($ditem['media']) {
219                $date = @filemtime(mediaFN($id));
220            } elseif (@file_exists(wikiFN($id))) {
221                $date = @filemtime(wikiFN($id));
222            } elseif($meta['date']['modified']) {
223                $date = $meta['date']['modified'];
224            } else {
225                $date = 0;
226            }
227            if($date) $item->date = date('r', $date);
228
229            // add title
230            if($conf['useheading'] && $meta['title']) {
231                $item->title = $meta['title'];
232            } else {
233                $item->title = $ditem['id'];
234            }
235            if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
236                $item->title .= ' - '.strip_tags($ditem['sum']);
237            }
238
239            // add item link
240            switch($opt['link_to']) {
241                case 'page':
242                    if($ditem['media']) {
243                        $item->link = media_managerURL(
244                            array(
245                                 'image' => $id,
246                                 'ns'    => getNS($id),
247                                 'rev'   => $date
248                            ), '&', true
249                        );
250                    } else {
251                        $item->link = wl($id, 'rev='.$date, true, '&');
252                    }
253                    break;
254                case 'rev':
255                    if($ditem['media']) {
256                        $item->link = media_managerURL(
257                            array(
258                                 'image'       => $id,
259                                 'ns'          => getNS($id),
260                                 'rev'         => $date,
261                                 'tab_details' => 'history'
262                            ), '&', true
263                        );
264                    } else {
265                        $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
266                    }
267                    break;
268                case 'current':
269                    if($ditem['media']) {
270                        $item->link = media_managerURL(
271                            array(
272                                 'image' => $id,
273                                 'ns'    => getNS($id)
274                            ), '&', true
275                        );
276                    } else {
277                        $item->link = wl($id, '', true, '&');
278                    }
279                    break;
280                case 'diff':
281                default:
282                    if($ditem['media']) {
283                        $item->link = media_managerURL(
284                            array(
285                                 'image'       => $id,
286                                 'ns'          => getNS($id),
287                                 'rev'         => $date,
288                                 'tab_details' => 'history',
289                                 'mediado'     => 'diff'
290                            ), '&', true
291                        );
292                    } else {
293                        $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
294                    }
295            }
296
297            // add item content
298            switch($opt['item_content']) {
299                case 'diff':
300                case 'htmldiff':
301                    if($ditem['media']) {
302                        $medialog = new MediaChangeLog($id);
303                        $revs  = $medialog->getRevisions(0, 1);
304                        $rev   = $revs[0];
305                        $src_r = '';
306                        $src_l = '';
307
308                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)), 300)) {
309                            $more  = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
310                            $src_r = ml($id, $more, true, '&amp;', true);
311                        }
312                        if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
313                            $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
314                            $src_l = ml($id, $more, true, '&amp;', true);
315                        }
316                        $content = '';
317                        if($src_r) {
318                            $content = '<table>';
319                            $content .= '<tr><th width="50%">'.$rev.'</th>';
320                            $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
321                            $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
322                            $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
323                            $content .= '</table>';
324                        }
325
326                    } else {
327                        require_once(DOKU_INC.'inc/DifferenceEngine.php');
328                        $pagelog = new PageChangeLog($id);
329                        $revs = $pagelog->getRevisions(0, 1);
330                        $rev  = $revs[0];
331
332                        if($rev) {
333                            $df = new Diff(explode("\n", rawWiki($id, $rev)),
334                                           explode("\n", rawWiki($id, '')));
335                        } else {
336                            $df = new Diff(array(''),
337                                           explode("\n", rawWiki($id, '')));
338                        }
339
340                        if($opt['item_content'] == 'htmldiff') {
341                            // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
342                            $tdf     = new TableDiffFormatter();
343                            $content = '<table>';
344                            $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
345                            $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
346                            $content .= $tdf->format($df);
347                            $content .= '</table>';
348                        } else {
349                            // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
350                            $udf     = new UnifiedDiffFormatter();
351                            $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
352                        }
353                    }
354                    break;
355                case 'html':
356                    if($ditem['media']) {
357                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
358                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
359                            $src     = ml($id, $more, true, '&amp;', true);
360                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
361                        } else {
362                            $content = '';
363                        }
364                    } else {
365                        if (@filemtime(wikiFN($id)) === $date) {
366                            $content = p_wiki_xhtml($id, '', false);
367                        } else {
368                            $content = p_wiki_xhtml($id, $date, false);
369                        }
370                        // no TOC in feeds
371                        $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
372
373                        // add alignment for images
374                        $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
375                        $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
376
377                        // make URLs work when canonical is not set, regexp instead of rerendering!
378                        if(!$conf['canonical']) {
379                            $base    = preg_quote(DOKU_REL, '/');
380                            $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
381                        }
382                    }
383
384                    break;
385                case 'abstract':
386                default:
387                    if($ditem['media']) {
388                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
389                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
390                            $src     = ml($id, $more, true, '&amp;', true);
391                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
392                        } else {
393                            $content = '';
394                        }
395                    } else {
396                        $content = $meta['description']['abstract'];
397                    }
398            }
399            $item->description = $content; //FIXME a plugin hook here could be senseful
400
401            // add user
402            # FIXME should the user be pulled from metadata as well?
403            $user         = @$ditem['user']; // the @ spares time repeating lookup
404            $item->author = '';
405            if($user && $conf['useacl'] && $auth) {
406                $userInfo = $auth->getUserData($user);
407                if($userInfo) {
408                    switch($conf['showuseras']) {
409                        case 'username':
410                            $item->author = $userInfo['name'];
411                            break;
412                        default:
413                            $item->author = $user;
414                            break;
415                    }
416                } else {
417                    $item->author = $user;
418                }
419                if($userInfo && !$opt['guardmail']) {
420                    $item->authorEmail = $userInfo['mail'];
421                } else {
422                    //cannot obfuscate because some RSS readers may check validity
423                    $item->authorEmail = $user.'@'.$ditem['ip'];
424                }
425            } elseif($user) {
426                // this happens when no ACL but some Apache auth is used
427                $item->author      = $user;
428                $item->authorEmail = $user.'@'.$ditem['ip'];
429            } else {
430                $item->authorEmail = 'anonymous@'.$ditem['ip'];
431            }
432
433            // add category
434            if(isset($meta['subject'])) {
435                $item->category = $meta['subject'];
436            } else {
437                $cat = getNS($id);
438                if($cat) $item->category = $cat;
439            }
440
441            // finally add the item to the feed object, after handing it to registered plugins
442            $evdata = array(
443                'item'  => &$item,
444                'opt'   => &$opt,
445                'ditem' => &$ditem,
446                'rss'   => &$rss
447            );
448            $evt    = new Doku_Event('FEED_ITEM_ADD', $evdata);
449            if($evt->advise_before()) {
450                $rss->addItem($item);
451            }
452            $evt->advise_after(); // for completeness
453        }
454    }
455    $event->advise_after();
456}
457
458/**
459 * Add recent changed pages to the feed object
460 *
461 * @author Andreas Gohr <andi@splitbrain.org>
462 */
463function rssRecentChanges($opt) {
464    global $conf;
465    $flags = RECENTS_SKIP_DELETED;
466    if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
467    if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
468    if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
469
470    $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
471    return $recents;
472}
473
474/**
475 * Add all pages of a namespace to the feed object
476 *
477 * @author Andreas Gohr <andi@splitbrain.org>
478 */
479function rssListNamespace($opt) {
480    require_once(DOKU_INC.'inc/search.php');
481    global $conf;
482
483    $ns = ':'.cleanID($opt['namespace']);
484    $ns = str_replace(':', '/', $ns);
485
486    $data = array();
487    $search_opts = array(
488        'depth' => 1,
489        'pagesonly' => true,
490        'listfiles' => true
491    );
492    search($data, $conf['datadir'], 'search_universal', $search_opts, $ns);
493
494    return $data;
495}
496
497/**
498 * Add the result of a full text search to the feed object
499 *
500 * @author Andreas Gohr <andi@splitbrain.org>
501 */
502function rssSearch($opt) {
503    if(!$opt['search_query']) return array();
504
505    require_once(DOKU_INC.'inc/fulltext.php');
506    $data = ft_pageSearch($opt['search_query'], $poswords);
507    $data = array_keys($data);
508
509    return $data;
510}
511
512//Setup VIM: ex: et ts=4 :
513