xref: /dokuwiki/feed.php (revision 15f699ac4db38c7098b4ae4cd0782dff13d46637) 
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 $INPUT
10 */
11
12use dokuwiki\Cache\Cache;
13use dokuwiki\ChangeLog\MediaChangeLog;
14use dokuwiki\ChangeLog\PageChangeLog;
15use dokuwiki\Extension\AuthPlugin;
16use dokuwiki\Extension\Event;
17use dokuwiki\Search\FulltextSearch;
18
19if(!defined('DOKU_INC')) define('DOKU_INC', dirname(__FILE__).'/');
20require_once(DOKU_INC.'inc/init.php');
21
22//close session
23session_write_close();
24
25//feed disabled?
26if(!actionOK('rss')) {
27    http_status(404);
28    echo '<error>RSS feed is disabled.</error>';
29    exit;
30}
31
32// get params
33$opt = rss_parseOptions();
34
35// the feed is dynamic - we need a cache for each combo
36// (but most people just use the default feed so it's still effective)
37$key   = join('', array_values($opt)).'$'.$_SERVER['REMOTE_USER'].'$'.$_SERVER['HTTP_HOST'].$_SERVER['SERVER_PORT'];
38$cache = new Cache($key, '.feed');
39
40// prepare cache depends
41$depends['files'] = getConfigFiles('main');
42$depends['age']   = $conf['rss_update'];
43$depends['purge'] = $INPUT->bool('purge');
44
45// check cacheage and deliver if nothing has changed since last
46// time or the update interval has not passed, also handles conditional requests
47header('Cache-Control: must-revalidate, post-check=0, pre-check=0');
48header('Pragma: public');
49header('Content-Type: application/xml; charset=utf-8');
50header('X-Robots-Tag: noindex');
51if($cache->useCache($depends)) {
52    http_conditionalRequest($cache->getTime());
53    if($conf['allowdebug']) header("X-CacheUsed: $cache->cache");
54    print $cache->retrieveCache();
55    exit;
56} else {
57    http_conditionalRequest(time());
58}
59
60// create new feed
61$rss                 = new UniversalFeedCreator();
62$rss->title          = $conf['title'].(($opt['namespace']) ? ' '.$opt['namespace'] : '');
63$rss->link           = DOKU_URL;
64$rss->syndicationURL = DOKU_URL.'feed.php';
65$rss->cssStyleSheet  = DOKU_URL.'lib/exe/css.php?s=feed';
66
67$image        = new FeedImage();
68$image->title = $conf['title'];
69$image->url   = tpl_getMediaFile(array(':wiki:favicon.ico', ':favicon.ico', 'images/favicon.ico'), true);
70$image->link  = DOKU_URL;
71$rss->image   = $image;
72
73$data  = null;
74$modes = array(
75    'list'   => 'rssListNamespace',
76    'search' => 'rssSearch',
77    'recent' => 'rssRecentChanges'
78);
79if(isset($modes[$opt['feed_mode']])) {
80    $data = $modes[$opt['feed_mode']]($opt);
81} else {
82    $eventData = array(
83        'opt'  => &$opt,
84        'data' => &$data,
85    );
86    $event     = new Event('FEED_MODE_UNKNOWN', $eventData);
87    if($event->advise_before(true)) {
88        echo sprintf('<error>Unknown feed mode %s</error>', hsc($opt['feed_mode']));
89        exit;
90    }
91    $event->advise_after();
92}
93
94rss_buildItems($rss, $data, $opt);
95$feed = $rss->createFeed($opt['feed_type']);
96
97// save cachefile
98$cache->storeCache($feed);
99
100// finally deliver
101print $feed;
102
103// ---------------------------------------------------------------- //
104
105/**
106 * Get URL parameters and config options and return an initialized option array
107 *
108 * @author Andreas Gohr <andi@splitbrain.org>
109 */
110function rss_parseOptions() {
111    global $conf;
112    global $INPUT;
113
114    $opt = array();
115
116    foreach(array(
117                // Basic feed properties
118                // Plugins may probably want to add new values to these
119                // properties for implementing own feeds
120
121                // One of: list, search, recent
122                'feed_mode'    => array('str', 'mode', 'recent'),
123                // One of: diff, page, rev, current
124                'link_to'      => array('str', 'linkto', $conf['rss_linkto']),
125                // One of: abstract, diff, htmldiff, html
126                'item_content' => array('str', 'content', $conf['rss_content']),
127
128                // Special feed properties
129                // These are only used by certain feed_modes
130
131                // String, used for feed title, in list and rc mode
132                'namespace'    => array('str', 'ns', null),
133                // Positive integer, only used in rc mode
134                'items'        => array('int', 'num', $conf['recent']),
135                // Boolean, only used in rc mode
136                'show_minor'   => array('bool', 'minor', false),
137                // Boolean, only used in rc mode
138                'only_new'     => array('bool', 'onlynewpages', false),
139                // String, only used in list mode
140                'sort'         => array('str', 'sort', 'natural'),
141                // String, only used in search mode
142                'search_query' => array('str', 'q', null),
143                // One of: pages, media, both
144                'content_type' => array('str', 'view', $conf['rss_media'])
145
146            ) as $name => $val) {
147        $opt[$name] = $INPUT->{$val[0]}($val[1], $val[2], true);
148    }
149
150    $opt['items']      = max(0, (int) $opt['items']);
151    $opt['show_minor'] = (bool) $opt['show_minor'];
152    $opt['only_new']   = (bool) $opt['only_new'];
153    $opt['sort'] = valid_input_set('sort', array('default' => 'natural', 'date'), $opt);
154
155    $opt['guardmail'] = ($conf['mailguard'] != '' && $conf['mailguard'] != 'none');
156
157    $type = $INPUT->valid(
158        'type',
159        array( 'rss', 'rss2', 'atom', 'atom1', 'rss1'),
160        $conf['rss_type']
161    );
162    switch($type) {
163        case 'rss':
164            $opt['feed_type'] = 'RSS0.91';
165            $opt['mime_type'] = 'text/xml';
166            break;
167        case 'rss2':
168            $opt['feed_type'] = 'RSS2.0';
169            $opt['mime_type'] = 'text/xml';
170            break;
171        case 'atom':
172            $opt['feed_type'] = 'ATOM0.3';
173            $opt['mime_type'] = 'application/xml';
174            break;
175        case 'atom1':
176            $opt['feed_type'] = 'ATOM1.0';
177            $opt['mime_type'] = 'application/atom+xml';
178            break;
179        default:
180            $opt['feed_type'] = 'RSS1.0';
181            $opt['mime_type'] = 'application/xml';
182    }
183
184    $eventData = array(
185        'opt' => &$opt,
186    );
187    Event::createAndTrigger('FEED_OPTS_POSTPROCESS', $eventData);
188    return $opt;
189}
190
191/**
192 * Add recent changed pages to a feed object
193 *
194 * @author Andreas Gohr <andi@splitbrain.org>
195 * @param  FeedCreator $rss the FeedCreator Object
196 * @param  array       $data the items to add
197 * @param  array       $opt  the feed options
198 */
199function rss_buildItems(&$rss, &$data, $opt) {
200    global $conf;
201    global $lang;
202    /* @var AuthPlugin $auth */
203    global $auth;
204
205    $eventData = array(
206        'rss'  => &$rss,
207        'data' => &$data,
208        'opt'  => &$opt,
209    );
210    $event     = new Event('FEED_DATA_PROCESS', $eventData);
211    if($event->advise_before(false)) {
212        foreach($data as $ditem) {
213            if(!is_array($ditem)) {
214                // not an array? then only a list of IDs was given
215                $ditem = array('id' => $ditem);
216            }
217
218            $item = new FeedItem();
219            $id   = $ditem['id'];
220            if(!$ditem['media']) {
221                $meta = p_get_metadata($id);
222            } else {
223                $meta = array();
224            }
225
226            // add date
227            if($ditem['date']) {
228                $date = $ditem['date'];
229            } elseif ($ditem['media']) {
230                $date = @filemtime(mediaFN($id));
231            } elseif (file_exists(wikiFN($id))) {
232                $date = @filemtime(wikiFN($id));
233            } elseif($meta['date']['modified']) {
234                $date = $meta['date']['modified'];
235            } else {
236                $date = 0;
237            }
238            if($date) $item->date = date('r', $date);
239
240            // add title
241            if($conf['useheading'] && $meta['title']) {
242                $item->title = $meta['title'];
243            } else {
244                $item->title = $ditem['id'];
245            }
246            if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
247                $item->title .= ' - '.strip_tags($ditem['sum']);
248            }
249
250            // add item link
251            switch($opt['link_to']) {
252                case 'page':
253                    if($ditem['media']) {
254                        $item->link = media_managerURL(
255                            array(
256                                 'image' => $id,
257                                 'ns'    => getNS($id),
258                                 'rev'   => $date
259                            ), '&', true
260                        );
261                    } else {
262                        $item->link = wl($id, 'rev='.$date, true, '&');
263                    }
264                    break;
265                case 'rev':
266                    if($ditem['media']) {
267                        $item->link = media_managerURL(
268                            array(
269                                 'image'       => $id,
270                                 'ns'          => getNS($id),
271                                 'rev'         => $date,
272                                 'tab_details' => 'history'
273                            ), '&', true
274                        );
275                    } else {
276                        $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
277                    }
278                    break;
279                case 'current':
280                    if($ditem['media']) {
281                        $item->link = media_managerURL(
282                            array(
283                                 'image' => $id,
284                                 'ns'    => getNS($id)
285                            ), '&', true
286                        );
287                    } else {
288                        $item->link = wl($id, '', true, '&');
289                    }
290                    break;
291                case 'diff':
292                default:
293                    if($ditem['media']) {
294                        $item->link = media_managerURL(
295                            array(
296                                 'image'       => $id,
297                                 'ns'          => getNS($id),
298                                 'rev'         => $date,
299                                 'tab_details' => 'history',
300                                 'mediado'     => 'diff'
301                            ), '&', true
302                        );
303                    } else {
304                        $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
305                    }
306            }
307
308            // add item content
309            switch($opt['item_content']) {
310                case 'diff':
311                case 'htmldiff':
312                    if($ditem['media']) {
313                        $medialog = new MediaChangeLog($id);
314                        $revs  = $medialog->getRevisions(0, 1);
315                        $rev   = $revs[0];
316                        $src_r = '';
317                        $src_l = '';
318
319                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)), 300)) {
320                            $more  = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
321                            $src_r = ml($id, $more, true, '&amp;', true);
322                        }
323                        if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
324                            $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
325                            $src_l = ml($id, $more, true, '&amp;', true);
326                        }
327                        $content = '';
328                        if($src_r) {
329                            $content = '<table>';
330                            $content .= '<tr><th width="50%">'.$rev.'</th>';
331                            $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
332                            $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
333                            $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
334                            $content .= '</table>';
335                        }
336
337                    } else {
338                        require_once(DOKU_INC.'inc/DifferenceEngine.php');
339                        $pagelog = new PageChangeLog($id);
340                        $revs = $pagelog->getRevisions(0, 1);
341                        $rev  = $revs[0];
342
343                        if($rev) {
344                            $df = new Diff(explode("\n", rawWiki($id, $rev)),
345                                           explode("\n", rawWiki($id, '')));
346                        } else {
347                            $df = new Diff(array(''),
348                                           explode("\n", rawWiki($id, '')));
349                        }
350
351                        if($opt['item_content'] == 'htmldiff') {
352                            // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
353                            $tdf     = new TableDiffFormatter();
354                            $content = '<table>';
355                            $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
356                            $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
357                            $content .= $tdf->format($df);
358                            $content .= '</table>';
359                        } else {
360                            // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
361                            $udf     = new UnifiedDiffFormatter();
362                            $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
363                        }
364                    }
365                    break;
366                case 'html':
367                    if($ditem['media']) {
368                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
369                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
370                            $src     = ml($id, $more, true, '&amp;', true);
371                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
372                        } else {
373                            $content = '';
374                        }
375                    } else {
376                        if (@filemtime(wikiFN($id)) === $date) {
377                            $content = p_wiki_xhtml($id, '', false);
378                        } else {
379                            $content = p_wiki_xhtml($id, $date, false);
380                        }
381                        // no TOC in feeds
382                        $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
383
384                        // add alignment for images
385                        $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
386                        $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
387
388                        // make URLs work when canonical is not set, regexp instead of rerendering!
389                        if(!$conf['canonical']) {
390                            $base    = preg_quote(DOKU_REL, '/');
391                            $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
392                        }
393                    }
394
395                    break;
396                case 'abstract':
397                default:
398                    if($ditem['media']) {
399                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
400                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
401                            $src     = ml($id, $more, true, '&amp;', true);
402                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
403                        } else {
404                            $content = '';
405                        }
406                    } else {
407                        $content = $meta['description']['abstract'];
408                    }
409            }
410            $item->description = $content; //FIXME a plugin hook here could be senseful
411
412            // add user
413            # FIXME should the user be pulled from metadata as well?
414            $user         = @$ditem['user']; // the @ spares time repeating lookup
415            if(blank($user)) {
416                $item->author = 'Anonymous';
417                $item->authorEmail = 'anonymous@undisclosed.example.com';
418            } else {
419                $item->author = $user;
420                $item->authorEmail = $user . '@undisclosed.example.com';
421
422                // get real user name if configured
423                if($conf['useacl'] && $auth) {
424                    $userInfo = $auth->getUserData($user);
425                    if($userInfo) {
426                        switch($conf['showuseras']) {
427                            case 'username':
428                            case 'username_link':
429                                $item->author = $userInfo['name'];
430                                break;
431                            default:
432                                $item->author = $user;
433                                break;
434                        }
435                    } else {
436                        $item->author = $user;
437                    }
438                }
439            }
440
441            // add category
442            if(isset($meta['subject'])) {
443                $item->category = $meta['subject'];
444            } else {
445                $cat = getNS($id);
446                if($cat) $item->category = $cat;
447            }
448
449            // finally add the item to the feed object, after handing it to registered plugins
450            $evdata = array(
451                'item'  => &$item,
452                'opt'   => &$opt,
453                'ditem' => &$ditem,
454                'rss'   => &$rss
455            );
456            $evt    = new Event('FEED_ITEM_ADD', $evdata);
457            if($evt->advise_before()) {
458                $rss->addItem($item);
459            }
460            $evt->advise_after(); // for completeness
461        }
462    }
463    $event->advise_after();
464}
465
466/**
467 * Add recent changed pages to the feed object
468 *
469 * @author Andreas Gohr <andi@splitbrain.org>
470 */
471function rssRecentChanges($opt) {
472    global $conf;
473    $flags = 0;
474    if(!$conf['rss_show_deleted']) $flags += RECENTS_SKIP_DELETED;
475    if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
476    if($opt['only_new']) $flags += RECENTS_ONLY_CREATION;
477    if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
478    if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
479
480    $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
481    return $recents;
482}
483
484/**
485 * Add all pages of a namespace to the feed object
486 *
487 * @author Andreas Gohr <andi@splitbrain.org>
488 */
489function rssListNamespace($opt) {
490    require_once(DOKU_INC.'inc/search.php');
491    global $conf;
492
493    $ns = ':'.cleanID($opt['namespace']);
494    $ns = utf8_encodeFN(str_replace(':', '/', $ns));
495
496    $data = array();
497    $search_opts = array(
498        'depth' => 1,
499        'pagesonly' => true,
500        'listfiles' => true
501    );
502    search($data, $conf['datadir'], 'search_universal', $search_opts, $ns, $lvl = 1, $opt['sort']);
503
504    return $data;
505}
506
507/**
508 * Add the result of a full text search to the feed object
509 *
510 * @author Andreas Gohr <andi@splitbrain.org>
511 */
512function rssSearch($opt) {
513    if (!$opt['search_query']) return array();
514
515    $FulltextSearch = FulltextSearch::getInstance();
516    $data = $FulltextSearch->pageSearch($opt['search_query'], $poswords);
517    $data = array_keys($data);
518
519    return $data;
520}
521
522//Setup VIM: ex: et ts=4 :
523