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                // Boolean, only used in rc mode
137                'only_new'     => array('bool', 'onlynewpages', false),
138                // String, only used in list mode
139                'sort'         => array('str', 'sort', 'natural'),
140                // String, only used in search mode
141                'search_query' => array('str', 'q', null),
142                // One of: pages, media, both
143                'content_type' => array('str', 'view', $conf['rss_media'])
144
145            ) as $name => $val) {
146        $opt[$name] = $INPUT->{$val[0]}($val[1], $val[2], true);
147    }
148
149    $opt['items']      = max(0, (int) $opt['items']);
150    $opt['show_minor'] = (bool) $opt['show_minor'];
151    $opt['only_new']   = (bool) $opt['only_new'];
152    $opt['sort'] = valid_input_set('sort', array('default' => 'natural', 'date'), $opt);
153
154    $opt['guardmail'] = ($conf['mailguard'] != '' && $conf['mailguard'] != 'none');
155
156    $type = $INPUT->valid(
157        'type',
158        array( 'rss', 'rss2', 'atom', 'atom1', 'rss1'),
159        $conf['rss_type']
160    );
161    switch($type) {
162        case 'rss':
163            $opt['feed_type'] = 'RSS0.91';
164            $opt['mime_type'] = 'text/xml';
165            break;
166        case 'rss2':
167            $opt['feed_type'] = 'RSS2.0';
168            $opt['mime_type'] = 'text/xml';
169            break;
170        case 'atom':
171            $opt['feed_type'] = 'ATOM0.3';
172            $opt['mime_type'] = 'application/xml';
173            break;
174        case 'atom1':
175            $opt['feed_type'] = 'ATOM1.0';
176            $opt['mime_type'] = 'application/atom+xml';
177            break;
178        default:
179            $opt['feed_type'] = 'RSS1.0';
180            $opt['mime_type'] = 'application/xml';
181    }
182
183    $eventData = array(
184        'opt' => &$opt,
185    );
186    Event::createAndTrigger('FEED_OPTS_POSTPROCESS', $eventData);
187    return $opt;
188}
189
190/**
191 * Add recent changed pages to a feed object
192 *
193 * @author Andreas Gohr <andi@splitbrain.org>
194 * @param  FeedCreator $rss the FeedCreator Object
195 * @param  array       $data the items to add
196 * @param  array       $opt  the feed options
197 */
198function rss_buildItems(&$rss, &$data, $opt) {
199    global $conf;
200    global $lang;
201    /* @var AuthPlugin $auth */
202    global $auth;
203
204    $eventData = array(
205        'rss'  => &$rss,
206        'data' => &$data,
207        'opt'  => &$opt,
208    );
209    $event     = new Event('FEED_DATA_PROCESS', $eventData);
210    if($event->advise_before(false)) {
211        foreach($data as $ditem) {
212            if(!is_array($ditem)) {
213                // not an array? then only a list of IDs was given
214                $ditem = array('id' => $ditem);
215            }
216
217            $item = new FeedItem();
218            $id   = $ditem['id'];
219            if(!$ditem['media']) {
220                $meta = p_get_metadata($id);
221            } else {
222                $meta = array();
223            }
224
225            // add date
226            if($ditem['date']) {
227                $date = $ditem['date'];
228            } elseif ($ditem['media']) {
229                $date = @filemtime(mediaFN($id));
230            } elseif (file_exists(wikiFN($id))) {
231                $date = @filemtime(wikiFN($id));
232            } elseif($meta['date']['modified']) {
233                $date = $meta['date']['modified'];
234            } else {
235                $date = 0;
236            }
237            if($date) $item->date = date('r', $date);
238
239            // add title
240            if($conf['useheading'] && $meta['title']) {
241                $item->title = $meta['title'];
242            } else {
243                $item->title = $ditem['id'];
244            }
245            if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
246                $item->title .= ' - '.strip_tags($ditem['sum']);
247            }
248
249            // add item link
250            switch($opt['link_to']) {
251                case 'page':
252                    if($ditem['media']) {
253                        $item->link = media_managerURL(
254                            array(
255                                 'image' => $id,
256                                 'ns'    => getNS($id),
257                                 'rev'   => $date
258                            ), '&', true
259                        );
260                    } else {
261                        $item->link = wl($id, 'rev='.$date, true, '&');
262                    }
263                    break;
264                case 'rev':
265                    if($ditem['media']) {
266                        $item->link = media_managerURL(
267                            array(
268                                 'image'       => $id,
269                                 'ns'          => getNS($id),
270                                 'rev'         => $date,
271                                 'tab_details' => 'history'
272                            ), '&', true
273                        );
274                    } else {
275                        $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
276                    }
277                    break;
278                case 'current':
279                    if($ditem['media']) {
280                        $item->link = media_managerURL(
281                            array(
282                                 'image' => $id,
283                                 'ns'    => getNS($id)
284                            ), '&', true
285                        );
286                    } else {
287                        $item->link = wl($id, '', true, '&');
288                    }
289                    break;
290                case 'diff':
291                default:
292                    if($ditem['media']) {
293                        $item->link = media_managerURL(
294                            array(
295                                 'image'       => $id,
296                                 'ns'          => getNS($id),
297                                 'rev'         => $date,
298                                 'tab_details' => 'history',
299                                 'mediado'     => 'diff'
300                            ), '&', true
301                        );
302                    } else {
303                        $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
304                    }
305            }
306
307            // add item content
308            switch($opt['item_content']) {
309                case 'diff':
310                case 'htmldiff':
311                    if($ditem['media']) {
312                        $medialog = new MediaChangeLog($id);
313                        $revs  = $medialog->getRevisions(0, 1);
314                        $rev   = $revs[0];
315                        $src_r = '';
316                        $src_l = '';
317
318                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)), 300)) {
319                            $more  = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
320                            $src_r = ml($id, $more, true, '&amp;', true);
321                        }
322                        if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
323                            $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
324                            $src_l = ml($id, $more, true, '&amp;', true);
325                        }
326                        $content = '';
327                        if($src_r) {
328                            $content = '<table>';
329                            $content .= '<tr><th width="50%">'.$rev.'</th>';
330                            $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
331                            $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
332                            $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
333                            $content .= '</table>';
334                        }
335
336                    } else {
337                        require_once(DOKU_INC.'inc/DifferenceEngine.php');
338                        $pagelog = new PageChangeLog($id);
339                        $revs = $pagelog->getRevisions(0, 1);
340                        $rev  = $revs[0];
341
342                        if($rev) {
343                            $df = new Diff(explode("\n", rawWiki($id, $rev)),
344                                           explode("\n", rawWiki($id, '')));
345                        } else {
346                            $df = new Diff(array(''),
347                                           explode("\n", rawWiki($id, '')));
348                        }
349
350                        if($opt['item_content'] == 'htmldiff') {
351                            // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
352                            $tdf     = new TableDiffFormatter();
353                            $content = '<table>';
354                            $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
355                            $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
356                            $content .= $tdf->format($df);
357                            $content .= '</table>';
358                        } else {
359                            // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
360                            $udf     = new UnifiedDiffFormatter();
361                            $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
362                        }
363                    }
364                    break;
365                case 'html':
366                    if($ditem['media']) {
367                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
368                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
369                            $src     = ml($id, $more, true, '&amp;', true);
370                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
371                        } else {
372                            $content = '';
373                        }
374                    } else {
375                        if (@filemtime(wikiFN($id)) === $date) {
376                            $content = p_wiki_xhtml($id, '', false);
377                        } else {
378                            $content = p_wiki_xhtml($id, $date, false);
379                        }
380                        // no TOC in feeds
381                        $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
382
383                        // add alignment for images
384                        $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
385                        $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
386
387                        // make URLs work when canonical is not set, regexp instead of rerendering!
388                        if(!$conf['canonical']) {
389                            $base    = preg_quote(DOKU_REL, '/');
390                            $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
391                        }
392                    }
393
394                    break;
395                case 'abstract':
396                default:
397                    if($ditem['media']) {
398                        if($size = media_image_preview_size($id, '', new JpegMeta(mediaFN($id)))) {
399                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
400                            $src     = ml($id, $more, true, '&amp;', true);
401                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
402                        } else {
403                            $content = '';
404                        }
405                    } else {
406                        $content = $meta['description']['abstract'];
407                    }
408            }
409            $item->description = $content; //FIXME a plugin hook here could be senseful
410
411            // add user
412            # FIXME should the user be pulled from metadata as well?
413            $user         = @$ditem['user']; // the @ spares time repeating lookup
414            if(blank($user)) {
415                $item->author = 'Anonymous';
416                $item->authorEmail = 'anonymous@undisclosed.example.com';
417            } else {
418                $item->author = $user;
419                $item->authorEmail = $user . '@undisclosed.example.com';
420
421                // get real user name if configured
422                if($conf['useacl'] && $auth) {
423                    $userInfo = $auth->getUserData($user);
424                    if($userInfo) {
425                        switch($conf['showuseras']) {
426                            case 'username':
427                            case 'username_link':
428                                $item->author = $userInfo['name'];
429                                break;
430                            default:
431                                $item->author = $user;
432                                break;
433                        }
434                    } else {
435                        $item->author = $user;
436                    }
437                }
438            }
439
440            // add category
441            if(isset($meta['subject'])) {
442                $item->category = $meta['subject'];
443            } else {
444                $cat = getNS($id);
445                if($cat) $item->category = $cat;
446            }
447
448            // finally add the item to the feed object, after handing it to registered plugins
449            $evdata = array(
450                'item'  => &$item,
451                'opt'   => &$opt,
452                'ditem' => &$ditem,
453                'rss'   => &$rss
454            );
455            $evt    = new Event('FEED_ITEM_ADD', $evdata);
456            if($evt->advise_before()) {
457                $rss->addItem($item);
458            }
459            $evt->advise_after(); // for completeness
460        }
461    }
462    $event->advise_after();
463}
464
465/**
466 * Add recent changed pages to the feed object
467 *
468 * @author Andreas Gohr <andi@splitbrain.org>
469 */
470function rssRecentChanges($opt) {
471    global $conf;
472    $flags = 0;
473    if(!$conf['rss_show_deleted']) $flags += RECENTS_SKIP_DELETED;
474    if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
475    if($opt['only_new']) $flags += RECENTS_ONLY_CREATION;
476    if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
477    if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
478
479    $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
480    return $recents;
481}
482
483/**
484 * Add all pages of a namespace to the feed object
485 *
486 * @author Andreas Gohr <andi@splitbrain.org>
487 */
488function rssListNamespace($opt) {
489    require_once(DOKU_INC.'inc/search.php');
490    global $conf;
491
492    $ns = ':'.cleanID($opt['namespace']);
493    $ns = utf8_encodeFN(str_replace(':', '/', $ns));
494
495    $data = array();
496    $search_opts = array(
497        'depth' => 1,
498        'pagesonly' => true,
499        'listfiles' => true
500    );
501    search($data, $conf['datadir'], 'search_universal', $search_opts, $ns, $lvl = 1, $opt['sort']);
502
503    return $data;
504}
505
506/**
507 * Add the result of a full text search to the feed object
508 *
509 * @author Andreas Gohr <andi@splitbrain.org>
510 */
511function rssSearch($opt) {
512    if(!$opt['search_query']) return array();
513
514    require_once(DOKU_INC.'inc/fulltext.php');
515    $data = ft_pageSearch($opt['search_query'], $poswords);
516    $data = array_keys($data);
517
518    return $data;
519}
520
521//Setup VIM: ex: et ts=4 :
522