xref: /dokuwiki/feed.php (revision 7b299f22af679ba69fe0addbd6aaa8b37deffbac) 
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 $INPUT
10 */
11
12use dokuwiki\Cache\Cache;
13use dokuwiki\ChangeLog\MediaChangeLog;
14use dokuwiki\ChangeLog\PageChangeLog;
15use dokuwiki\Extension\AuthPlugin;
16use dokuwiki\Extension\Event;
17
18if(!defined('DOKU_INC')) define('DOKU_INC', dirname(__FILE__).'/');
19require_once(DOKU_INC.'inc/init.php');
20
21//close session
22session_write_close();
23
24//feed disabled?
25if(!actionOK('rss')) {
26    http_status(404);
27    echo '<error>RSS feed is disabled.</error>';
28    exit;
29}
30
31// get params
32$opt = rss_parseOptions();
33
34// the feed is dynamic - we need a cache for each combo
35// (but most people just use the default feed so it's still effective)
36$key   = join('', array_values($opt)).'$'.$_SERVER['REMOTE_USER'].'$'.$_SERVER['HTTP_HOST'].$_SERVER['SERVER_PORT'];
37$cache = new Cache($key, '.feed');
38
39// prepare cache depends
40$depends['files'] = getConfigFiles('main');
41$depends['age']   = $conf['rss_update'];
42$depends['purge'] = $INPUT->bool('purge');
43
44// check cacheage and deliver if nothing has changed since last
45// time or the update interval has not passed, also handles conditional requests
46header('Cache-Control: must-revalidate, post-check=0, pre-check=0');
47header('Pragma: public');
48header('Content-Type: application/xml; charset=utf-8');
49header('X-Robots-Tag: noindex');
50if($cache->useCache($depends)) {
51    http_conditionalRequest($cache->getTime());
52    if($conf['allowdebug']) header("X-CacheUsed: $cache->cache");
53    print $cache->retrieveCache();
54    exit;
55} else {
56    http_conditionalRequest(time());
57}
58
59// create new feed
60$rss                 = new UniversalFeedCreator();
61$rss->title          = $conf['title'].(($opt['namespace']) ? ' '.$opt['namespace'] : '');
62$rss->link           = DOKU_URL;
63$rss->syndicationURL = DOKU_URL.'feed.php';
64$rss->cssStyleSheet  = DOKU_URL.'lib/exe/css.php?s=feed';
65
66$image        = new FeedImage();
67$image->title = $conf['title'];
68$image->url   = tpl_getMediaFile(array(':wiki:favicon.ico', ':favicon.ico', 'images/favicon.ico'), true);
69$image->link  = DOKU_URL;
70$rss->image   = $image;
71
72$data  = null;
73$modes = array(
74    'list'   => 'rssListNamespace',
75    'search' => 'rssSearch',
76    'recent' => 'rssRecentChanges'
77);
78if(isset($modes[$opt['feed_mode']])) {
79    $data = $modes[$opt['feed_mode']]($opt);
80} else {
81    $eventData = array(
82        'opt'  => &$opt,
83        'data' => &$data,
84    );
85    $event     = new Event('FEED_MODE_UNKNOWN', $eventData);
86    if($event->advise_before(true)) {
87        echo sprintf('<error>Unknown feed mode %s</error>', hsc($opt['feed_mode']));
88        exit;
89    }
90    $event->advise_after();
91}
92
93rss_buildItems($rss, $data, $opt);
94$feed = $rss->createFeed($opt['feed_type']);
95
96// save cachefile
97$cache->storeCache($feed);
98
99// finally deliver
100print $feed;
101
102// ---------------------------------------------------------------- //
103
104/**
105 * Get URL parameters and config options and return an initialized option array
106 *
107 * @author Andreas Gohr <andi@splitbrain.org>
108 */
109function rss_parseOptions() {
110    global $conf;
111    global $INPUT;
112
113    $opt = array();
114
115    foreach(array(
116                // Basic feed properties
117                // Plugins may probably want to add new values to these
118                // properties for implementing own feeds
119
120                // One of: list, search, recent
121                'feed_mode'    => array('str', 'mode', 'recent'),
122                // One of: diff, page, rev, current
123                'link_to'      => array('str', 'linkto', $conf['rss_linkto']),
124                // One of: abstract, diff, htmldiff, html
125                'item_content' => array('str', 'content', $conf['rss_content']),
126
127                // Special feed properties
128                // These are only used by certain feed_modes
129
130                // String, used for feed title, in list and rc mode
131                'namespace'    => array('str', 'ns', null),
132                // Positive integer, only used in rc mode
133                'items'        => array('int', 'num', $conf['recent']),
134                // Boolean, only used in rc mode
135                'show_minor'   => array('bool', 'minor', false),
136                // String, only used in list mode
137                'sort'         => array('str', 'sort', 'natural'),
138                // String, only used in search mode
139                'search_query' => array('str', 'q', null),
140                // One of: pages, media, both
141                'content_type' => array('str', 'view', $conf['rss_media'])
142
143            ) as $name => $val) {
144        $opt[$name] = $INPUT->{$val[0]}($val[1], $val[2], true);
145    }
146
147    $opt['items']      = max(0, (int) $opt['items']);
148    $opt['show_minor'] = (bool) $opt['show_minor'];
149    $opt['sort'] = valid_input_set('sort', array('default' => 'natural', 'date'), $opt);
150
151    $opt['guardmail'] = ($conf['mailguard'] != '' && $conf['mailguard'] != 'none');
152
153    $type = $INPUT->valid(
154        'type',
155        array( 'rss', 'rss2', 'atom', 'atom1', 'rss1'),
156        $conf['rss_type']
157    );
158    switch($type) {
159        case 'rss':
160            $opt['feed_type'] = 'RSS0.91';
161            $opt['mime_type'] = 'text/xml';
162            break;
163        case 'rss2':
164            $opt['feed_type'] = 'RSS2.0';
165            $opt['mime_type'] = 'text/xml';
166            break;
167        case 'atom':
168            $opt['feed_type'] = 'ATOM0.3';
169            $opt['mime_type'] = 'application/xml';
170            break;
171        case 'atom1':
172            $opt['feed_type'] = 'ATOM1.0';
173            $opt['mime_type'] = 'application/atom+xml';
174            break;
175        default:
176            $opt['feed_type'] = 'RSS1.0';
177            $opt['mime_type'] = 'application/xml';
178    }
179
180    $eventData = array(
181        'opt' => &$opt,
182    );
183    Event::createAndTrigger('FEED_OPTS_POSTPROCESS', $eventData);
184    return $opt;
185}
186
187/**
188 * Add recent changed pages to a feed object
189 *
190 * @author Andreas Gohr <andi@splitbrain.org>
191 * @param  FeedCreator $rss the FeedCreator Object
192 * @param  array       $data the items to add
193 * @param  array       $opt  the feed options
194 */
195function rss_buildItems(&$rss, &$data, $opt) {
196    global $conf;
197    global $lang;
198    /* @var AuthPlugin $auth */
199    global $auth;
200
201    $eventData = array(
202        'rss'  => &$rss,
203        'data' => &$data,
204        'opt'  => &$opt,
205    );
206    $event     = new Event('FEED_DATA_PROCESS', $eventData);
207    if($event->advise_before(false)) {
208        foreach($data as $ditem) {
209            if(!is_array($ditem)) {
210                // not an array? then only a list of IDs was given
211                $ditem = array('id' => $ditem);
212            }
213
214            $item = new FeedItem();
215            $id   = $ditem['id'];
216            if(!$ditem['media']) {
217                $meta = p_get_metadata($id);
218            } else {
219                $meta = array();
220            }
221
222            // add date
223            if($ditem['date']) {
224                $date = $ditem['date'];
225            } elseif ($ditem['media']) {
226                $date = @filemtime(mediaFN($id));
227            } elseif (file_exists(wikiFN($id))) {
228                $date = @filemtime(wikiFN($id));
229            } elseif($meta['date']['modified']) {
230                $date = $meta['date']['modified'];
231            } else {
232                $date = 0;
233            }
234            if($date) $item->date = date('r', $date);
235
236            // add title
237            if($conf['useheading'] && $meta['title']) {
238                $item->title = $meta['title'];
239            } else {
240                $item->title = $ditem['id'];
241            }
242            if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
243                $item->title .= ' - '.strip_tags($ditem['sum']);
244            }
245
246            // add item link
247            switch($opt['link_to']) {
248                case 'page':
249                    if($ditem['media']) {
250                        $item->link = media_managerURL(
251                            array(
252                                 'image' => $id,
253                                 'ns'    => getNS($id),
254                                 'rev'   => $date
255                            ), '&', true
256                        );
257                    } else {
258                        $item->link = wl($id, 'rev='.$date, true, '&');
259                    }
260                    break;
261                case 'rev':
262                    if($ditem['media']) {
263                        $item->link = media_managerURL(
264                            array(
265                                 'image'       => $id,
266                                 'ns'          => getNS($id),
267                                 'rev'         => $date,
268                                 'tab_details' => 'history'
269                            ), '&', true
270                        );
271                    } else {
272                        $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
273                    }
274                    break;
275                case 'current':
276                    if($ditem['media']) {
277                        $item->link = media_managerURL(
278                            array(
279                                 'image' => $id,
280                                 'ns'    => getNS($id)
281                            ), '&', true
282                        );
283                    } else {
284                        $item->link = wl($id, '', true, '&');
285                    }
286                    break;
287                case 'diff':
288                default:
289                    if($ditem['media']) {
290                        $item->link = media_managerURL(
291                            array(
292                                 'image'       => $id,
293                                 'ns'          => getNS($id),
294                                 'rev'         => $date,
295                                 'tab_details' => 'history',
296                                 'mediado'     => 'diff'
297                            ), '&', true
298                        );
299                    } else {
300                        $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
301                    }
302            }
303
304            // add item content
305            switch($opt['item_content']) {
306                case 'diff':
307                case 'htmldiff':
308                    if($ditem['media']) {
309                        $medialog = new MediaChangeLog($id);
310                        $revs  = $medialog->getRevisions(0, 1);
311                        $rev   = $revs[0];
312                        $src_r = '';
313                        $src_l = '';
314
315                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)), 300)) {
316                            $more  = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
317                            $src_r = ml($id, $more, true, '&amp;', true);
318                        }
319                        if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
320                            $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
321                            $src_l = ml($id, $more, true, '&amp;', true);
322                        }
323                        $content = '';
324                        if($src_r) {
325                            $content = '<table>';
326                            $content .= '<tr><th width="50%">'.$rev.'</th>';
327                            $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
328                            $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
329                            $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
330                            $content .= '</table>';
331                        }
332
333                    } else {
334                        require_once(DOKU_INC.'inc/DifferenceEngine.php');
335                        $pagelog = new PageChangeLog($id);
336                        $revs = $pagelog->getRevisions(0, 1);
337                        $rev  = $revs[0];
338
339                        if($rev) {
340                            $df = new Diff(explode("\n", rawWiki($id, $rev)),
341                                           explode("\n", rawWiki($id, '')));
342                        } else {
343                            $df = new Diff(array(''),
344                                           explode("\n", rawWiki($id, '')));
345                        }
346
347                        if($opt['item_content'] == 'htmldiff') {
348                            // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
349                            $tdf     = new TableDiffFormatter();
350                            $content = '<table>';
351                            $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
352                            $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
353                            $content .= $tdf->format($df);
354                            $content .= '</table>';
355                        } else {
356                            // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
357                            $udf     = new UnifiedDiffFormatter();
358                            $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
359                        }
360                    }
361                    break;
362                case 'html':
363                    if($ditem['media']) {
364                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
365                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
366                            $src     = ml($id, $more, true, '&amp;', true);
367                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
368                        } else {
369                            $content = '';
370                        }
371                    } else {
372                        if (@filemtime(wikiFN($id)) === $date) {
373                            $content = p_wiki_xhtml($id, '', false);
374                        } else {
375                            $content = p_wiki_xhtml($id, $date, false);
376                        }
377                        // no TOC in feeds
378                        $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
379
380                        // add alignment for images
381                        $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
382                        $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
383
384                        // make URLs work when canonical is not set, regexp instead of rerendering!
385                        if(!$conf['canonical']) {
386                            $base    = preg_quote(DOKU_REL, '/');
387                            $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
388                        }
389                    }
390
391                    break;
392                case 'abstract':
393                default:
394                    if($ditem['media']) {
395                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
396                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
397                            $src     = ml($id, $more, true, '&amp;', true);
398                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
399                        } else {
400                            $content = '';
401                        }
402                    } else {
403                        $content = $meta['description']['abstract'];
404                    }
405            }
406            $item->description = $content; //FIXME a plugin hook here could be senseful
407
408            // add user
409            # FIXME should the user be pulled from metadata as well?
410            $user         = @$ditem['user']; // the @ spares time repeating lookup
411            if(blank($user)) {
412                $item->author = 'Anonymous';
413                $item->authorEmail = 'anonymous@undisclosed.example.com';
414            } else {
415                $item->author = $user;
416                $item->authorEmail = $user . '@undisclosed.example.com';
417
418                // get real user name if configured
419                if($conf['useacl'] && $auth) {
420                    $userInfo = $auth->getUserData($user);
421                    if($userInfo) {
422                        switch($conf['showuseras']) {
423                            case 'username':
424                            case 'username_link':
425                                $item->author = $userInfo['name'];
426                                break;
427                            default:
428                                $item->author = $user;
429                                break;
430                        }
431                    } else {
432                        $item->author = $user;
433                    }
434                }
435            }
436
437            // add category
438            if(isset($meta['subject'])) {
439                $item->category = $meta['subject'];
440            } else {
441                $cat = getNS($id);
442                if($cat) $item->category = $cat;
443            }
444
445            // finally add the item to the feed object, after handing it to registered plugins
446            $evdata = array(
447                'item'  => &$item,
448                'opt'   => &$opt,
449                'ditem' => &$ditem,
450                'rss'   => &$rss
451            );
452            $evt    = new Event('FEED_ITEM_ADD', $evdata);
453            if($evt->advise_before()) {
454                $rss->addItem($item);
455            }
456            $evt->advise_after(); // for completeness
457        }
458    }
459    $event->advise_after();
460}
461
462/**
463 * Add recent changed pages to the feed object
464 *
465 * @author Andreas Gohr <andi@splitbrain.org>
466 */
467function rssRecentChanges($opt) {
468    global $conf;
469    //$flags = RECENTS_SKIP_DELETED;
470    if(!$conf['rss_show_deleted']) $flags = RECENTS_SKIP_DELETED;
471    if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
472    if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
473    if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
474
475    $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
476    return $recents;
477}
478
479/**
480 * Add all pages of a namespace to the feed object
481 *
482 * @author Andreas Gohr <andi@splitbrain.org>
483 */
484function rssListNamespace($opt) {
485    require_once(DOKU_INC.'inc/search.php');
486    global $conf;
487
488    $ns = ':'.cleanID($opt['namespace']);
489    $ns = utf8_encodeFN(str_replace(':', '/', $ns));
490
491    $data = array();
492    $search_opts = array(
493        'depth' => 1,
494        'pagesonly' => true,
495        'listfiles' => true
496    );
497    search($data, $conf['datadir'], 'search_universal', $search_opts, $ns, $lvl = 1, $opt['sort']);
498
499    return $data;
500}
501
502/**
503 * Add the result of a full text search to the feed object
504 *
505 * @author Andreas Gohr <andi@splitbrain.org>
506 */
507function rssSearch($opt) {
508    if(!$opt['search_query']) return array();
509
510    require_once(DOKU_INC.'inc/fulltext.php');
511    $data = ft_pageSearch($opt['search_query'], $poswords);
512    $data = array_keys($data);
513
514    return $data;
515}
516
517//Setup VIM: ex: et ts=4 :
518