xref: /dokuwiki/feed.php (revision 617a2b42a29fc3da7a22122792ebe2184ccbbc2f) 
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 UniversalFeedCreator();
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']);
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 list mode
131                'sort'         => array('str', 'sort', 'natural'),
132                // String, only used in search mode
133                'search_query' => array('str', 'q', null),
134                // One of: pages, media, both
135                'content_type' => array('str', 'view', $conf['rss_media'])
136
137            ) as $name => $val) {
138        $opt[$name] = $INPUT->{$val[0]}($val[1], $val[2], true);
139    }
140
141    $opt['items']      = max(0, (int) $opt['items']);
142    $opt['show_minor'] = (bool) $opt['show_minor'];
143    $opt['sort'] = valid_input_set('sort', array('default' => 'natural', 'date'), $opt);
144
145    $opt['guardmail'] = ($conf['mailguard'] != '' && $conf['mailguard'] != 'none');
146
147    $type = $INPUT->valid(
148        'type',
149        array( 'rss', 'rss2', 'atom', 'atom1', 'rss1'),
150        $conf['rss_type']
151    );
152    switch($type) {
153        case 'rss':
154            $opt['feed_type'] = 'RSS0.91';
155            $opt['mime_type'] = 'text/xml';
156            break;
157        case 'rss2':
158            $opt['feed_type'] = 'RSS2.0';
159            $opt['mime_type'] = 'text/xml';
160            break;
161        case 'atom':
162            $opt['feed_type'] = 'ATOM0.3';
163            $opt['mime_type'] = 'application/xml';
164            break;
165        case 'atom1':
166            $opt['feed_type'] = 'ATOM1.0';
167            $opt['mime_type'] = 'application/atom+xml';
168            break;
169        default:
170            $opt['feed_type'] = 'RSS1.0';
171            $opt['mime_type'] = 'application/xml';
172    }
173
174    $eventData = array(
175        'opt' => &$opt,
176    );
177    trigger_event('FEED_OPTS_POSTPROCESS', $eventData);
178    return $opt;
179}
180
181/**
182 * Add recent changed pages to a feed object
183 *
184 * @author Andreas Gohr <andi@splitbrain.org>
185 * @param  FeedCreator $rss the FeedCreator Object
186 * @param  array       $data the items to add
187 * @param  array       $opt  the feed options
188 */
189function rss_buildItems(&$rss, &$data, $opt) {
190    global $conf;
191    global $lang;
192    /* @var DokuWiki_Auth_Plugin $auth */
193    global $auth;
194
195    $eventData = array(
196        'rss'  => &$rss,
197        'data' => &$data,
198        'opt'  => &$opt,
199    );
200    $event     = new Doku_Event('FEED_DATA_PROCESS', $eventData);
201    if($event->advise_before(false)) {
202        foreach($data as $ditem) {
203            if(!is_array($ditem)) {
204                // not an array? then only a list of IDs was given
205                $ditem = array('id' => $ditem);
206            }
207
208            $item = new FeedItem();
209            $id   = $ditem['id'];
210            if(!$ditem['media']) {
211                $meta = p_get_metadata($id);
212            } else {
213                $meta = array();
214            }
215
216            // add date
217            if($ditem['date']) {
218                $date = $ditem['date'];
219            } elseif ($ditem['media']) {
220                $date = @filemtime(mediaFN($id));
221            } elseif (file_exists(wikiFN($id))) {
222                $date = @filemtime(wikiFN($id));
223            } elseif($meta['date']['modified']) {
224                $date = $meta['date']['modified'];
225            } else {
226                $date = 0;
227            }
228            if($date) $item->date = date('r', $date);
229
230            // add title
231            if($conf['useheading'] && $meta['title']) {
232                $item->title = $meta['title'];
233            } else {
234                $item->title = $ditem['id'];
235            }
236            if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
237                $item->title .= ' - '.strip_tags($ditem['sum']);
238            }
239
240            // add item link
241            switch($opt['link_to']) {
242                case 'page':
243                    if($ditem['media']) {
244                        $item->link = media_managerURL(
245                            array(
246                                 'image' => $id,
247                                 'ns'    => getNS($id),
248                                 'rev'   => $date
249                            ), '&', true
250                        );
251                    } else {
252                        $item->link = wl($id, 'rev='.$date, true, '&');
253                    }
254                    break;
255                case 'rev':
256                    if($ditem['media']) {
257                        $item->link = media_managerURL(
258                            array(
259                                 'image'       => $id,
260                                 'ns'          => getNS($id),
261                                 'rev'         => $date,
262                                 'tab_details' => 'history'
263                            ), '&', true
264                        );
265                    } else {
266                        $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
267                    }
268                    break;
269                case 'current':
270                    if($ditem['media']) {
271                        $item->link = media_managerURL(
272                            array(
273                                 'image' => $id,
274                                 'ns'    => getNS($id)
275                            ), '&', true
276                        );
277                    } else {
278                        $item->link = wl($id, '', true, '&');
279                    }
280                    break;
281                case 'diff':
282                default:
283                    if($ditem['media']) {
284                        $item->link = media_managerURL(
285                            array(
286                                 'image'       => $id,
287                                 'ns'          => getNS($id),
288                                 'rev'         => $date,
289                                 'tab_details' => 'history',
290                                 'mediado'     => 'diff'
291                            ), '&', true
292                        );
293                    } else {
294                        $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
295                    }
296            }
297
298            // add item content
299            switch($opt['item_content']) {
300                case 'diff':
301                case 'htmldiff':
302                    if($ditem['media']) {
303                        $medialog = new MediaChangeLog($id);
304                        $revs  = $medialog->getRevisions(0, 1);
305                        $rev   = $revs[0];
306                        $src_r = '';
307                        $src_l = '';
308
309                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)), 300)) {
310                            $more  = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
311                            $src_r = ml($id, $more, true, '&amp;', true);
312                        }
313                        if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
314                            $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
315                            $src_l = ml($id, $more, true, '&amp;', true);
316                        }
317                        $content = '';
318                        if($src_r) {
319                            $content = '<table>';
320                            $content .= '<tr><th width="50%">'.$rev.'</th>';
321                            $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
322                            $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
323                            $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
324                            $content .= '</table>';
325                        }
326
327                    } else {
328                        require_once(DOKU_INC.'inc/DifferenceEngine.php');
329                        $pagelog = new PageChangeLog($id);
330                        $revs = $pagelog->getRevisions(0, 1);
331                        $rev  = $revs[0];
332
333                        if($rev) {
334                            $df = new Diff(explode("\n", rawWiki($id, $rev)),
335                                           explode("\n", rawWiki($id, '')));
336                        } else {
337                            $df = new Diff(array(''),
338                                           explode("\n", rawWiki($id, '')));
339                        }
340
341                        if($opt['item_content'] == 'htmldiff') {
342                            // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
343                            $tdf     = new TableDiffFormatter();
344                            $content = '<table>';
345                            $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
346                            $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
347                            $content .= $tdf->format($df);
348                            $content .= '</table>';
349                        } else {
350                            // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
351                            $udf     = new UnifiedDiffFormatter();
352                            $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
353                        }
354                    }
355                    break;
356                case 'html':
357                    if($ditem['media']) {
358                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
359                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
360                            $src     = ml($id, $more, true, '&amp;', true);
361                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
362                        } else {
363                            $content = '';
364                        }
365                    } else {
366                        if (@filemtime(wikiFN($id)) === $date) {
367                            $content = p_wiki_xhtml($id, '', false);
368                        } else {
369                            $content = p_wiki_xhtml($id, $date, false);
370                        }
371                        // no TOC in feeds
372                        $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
373
374                        // add alignment for images
375                        $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
376                        $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
377
378                        // make URLs work when canonical is not set, regexp instead of rerendering!
379                        if(!$conf['canonical']) {
380                            $base    = preg_quote(DOKU_REL, '/');
381                            $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
382                        }
383                    }
384
385                    break;
386                case 'abstract':
387                default:
388                    if($ditem['media']) {
389                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
390                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
391                            $src     = ml($id, $more, true, '&amp;', true);
392                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
393                        } else {
394                            $content = '';
395                        }
396                    } else {
397                        $content = $meta['description']['abstract'];
398                    }
399            }
400            $item->description = $content; //FIXME a plugin hook here could be senseful
401
402            // add user
403            # FIXME should the user be pulled from metadata as well?
404            $user         = @$ditem['user']; // the @ spares time repeating lookup
405            $item->author = '';
406            if($user && $conf['useacl'] && $auth) {
407                $userInfo = $auth->getUserData($user);
408                if($userInfo) {
409                    switch($conf['showuseras']) {
410                        case 'username':
411                        case 'username_link':
412                            $item->author = $userInfo['name'];
413                            break;
414                        default:
415                            $item->author = $user;
416                            break;
417                    }
418                } else {
419                    $item->author = $user;
420                }
421                if($userInfo && !$opt['guardmail']) {
422                    $item->authorEmail = $userInfo['mail'];
423                } else {
424                    //cannot obfuscate because some RSS readers may check validity
425                    $item->authorEmail = $user.'@'.$ditem['ip'];
426                }
427            } elseif($user) {
428                // this happens when no ACL but some Apache auth is used
429                $item->author      = $user;
430                $item->authorEmail = $user.'@'.$ditem['ip'];
431            } else {
432                $item->authorEmail = 'anonymous@'.$ditem['ip'];
433            }
434
435            // add category
436            if(isset($meta['subject'])) {
437                $item->category = $meta['subject'];
438            } else {
439                $cat = getNS($id);
440                if($cat) $item->category = $cat;
441            }
442
443            // finally add the item to the feed object, after handing it to registered plugins
444            $evdata = array(
445                'item'  => &$item,
446                'opt'   => &$opt,
447                'ditem' => &$ditem,
448                'rss'   => &$rss
449            );
450            $evt    = new Doku_Event('FEED_ITEM_ADD', $evdata);
451            if($evt->advise_before()) {
452                $rss->addItem($item);
453            }
454            $evt->advise_after(); // for completeness
455        }
456    }
457    $event->advise_after();
458}
459
460/**
461 * Add recent changed pages to the feed object
462 *
463 * @author Andreas Gohr <andi@splitbrain.org>
464 */
465function rssRecentChanges($opt) {
466    global $conf;
467    $flags = RECENTS_SKIP_DELETED;
468    if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
469    if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
470    if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
471
472    $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
473    return $recents;
474}
475
476/**
477 * Add all pages of a namespace to the feed object
478 *
479 * @author Andreas Gohr <andi@splitbrain.org>
480 */
481function rssListNamespace($opt) {
482    require_once(DOKU_INC.'inc/search.php');
483    global $conf;
484
485    $ns = ':'.cleanID($opt['namespace']);
486    $ns = utf8_encodeFN(str_replace(':', '/', $ns));
487
488    $data = array();
489    $search_opts = array(
490        'depth' => 1,
491        'pagesonly' => true,
492        'listfiles' => true
493    );
494    search($data, $conf['datadir'], 'search_universal', $search_opts, $ns, $lvl = 1, $opt['sort']);
495
496    return $data;
497}
498
499/**
500 * Add the result of a full text search to the feed object
501 *
502 * @author Andreas Gohr <andi@splitbrain.org>
503 */
504function rssSearch($opt) {
505    if(!$opt['search_query']) return array();
506
507    require_once(DOKU_INC.'inc/fulltext.php');
508    $data = ft_pageSearch($opt['search_query'], $poswords);
509    $data = array_keys($data);
510
511    return $data;
512}
513
514//Setup VIM: ex: et ts=4 :
515