xref: /dokuwiki/feed.php (revision 477e59988f5e90600b6fdcda7c055ab8ea001e03)
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 $INPUT
10 */
11
12if(!defined('DOKU_INC')) define('DOKU_INC', dirname(__FILE__).'/');
13require_once(DOKU_INC.'inc/init.php');
14
15//close session
16session_write_close();
17
18// get params
19$opt = rss_parseOptions();
20
21// the feed is dynamic - we need a cache for each combo
22// (but most people just use the default feed so it's still effective)
23$key   = join('', array_values($opt)).'$'.$_SERVER['REMOTE_USER'].'$'.$_SERVER['HTTP_HOST'].$_SERVER['SERVER_PORT'];
24$cache = new cache($key, '.feed');
25
26// prepare cache depends
27$depends['files'] = getConfigFiles('main');
28$depends['age']   = $conf['rss_update'];
29$depends['purge'] = $INPUT->bool('purge');
30
31// check cacheage and deliver if nothing has changed since last
32// time or the update interval has not passed, also handles conditional requests
33header('Cache-Control: must-revalidate, post-check=0, pre-check=0');
34header('Pragma: public');
35header('Content-Type: application/xml; charset=utf-8');
36header('X-Robots-Tag: noindex');
37if($cache->useCache($depends)) {
38    http_conditionalRequest($cache->_time);
39    if($conf['allowdebug']) header("X-CacheUsed: $cache->cache");
40    print $cache->retrieveCache();
41    exit;
42} else {
43    http_conditionalRequest(time());
44}
45
46// create new feed
47$rss                 = new DokuWikiFeedCreator();
48$rss->title          = $conf['title'].(($opt['namespace']) ? ' '.$opt['namespace'] : '');
49$rss->link           = DOKU_URL;
50$rss->syndicationURL = DOKU_URL.'feed.php';
51$rss->cssStyleSheet  = DOKU_URL.'lib/exe/css.php?s=feed';
52
53$image        = new FeedImage();
54$image->title = $conf['title'];
55$image->url   = tpl_getMediaFile(array(':wiki:favicon.ico', ':favicon.ico', 'images/favicon.ico'), true);
56$image->link  = DOKU_URL;
57$rss->image   = $image;
58
59$data  = null;
60$modes = array(
61    'list'   => 'rssListNamespace',
62    'search' => 'rssSearch',
63    'recent' => 'rssRecentChanges'
64);
65if(isset($modes[$opt['feed_mode']])) {
66    $data = $modes[$opt['feed_mode']]($opt);
67} else {
68    $eventData = array(
69        'opt'  => &$opt,
70        'data' => &$data,
71    );
72    $event     = new Doku_Event('FEED_MODE_UNKNOWN', $eventData);
73    if($event->advise_before(true)) {
74        echo sprintf('<error>Unknown feed mode %s</error>', hsc($opt['feed_mode']));
75        exit;
76    }
77    $event->advise_after();
78}
79
80rss_buildItems($rss, $data, $opt);
81$feed = $rss->createFeed($opt['feed_type'], 'utf-8');
82
83// save cachefile
84$cache->storeCache($feed);
85
86// finally deliver
87print $feed;
88
89// ---------------------------------------------------------------- //
90
91/**
92 * Get URL parameters and config options and return an initialized option array
93 *
94 * @author Andreas Gohr <andi@splitbrain.org>
95 */
96function rss_parseOptions() {
97    global $conf;
98    global $INPUT;
99
100    $opt = array();
101
102    foreach(array(
103                // Basic feed properties
104                // Plugins may probably want to add new values to these
105                // properties for implementing own feeds
106
107                // One of: list, search, recent
108                'feed_mode'    => array('str', 'mode', 'recent'),
109                // One of: diff, page, rev, current
110                'link_to'      => array('str', 'linkto', $conf['rss_linkto']),
111                // One of: abstract, diff, htmldiff, html
112                'item_content' => array('str', 'content', $conf['rss_content']),
113
114                // Special feed properties
115                // These are only used by certain feed_modes
116
117                // String, used for feed title, in list and rc mode
118                'namespace'    => array('str', 'ns', null),
119                // Positive integer, only used in rc mode
120                'items'        => array('int', 'num', $conf['recent']),
121                // Boolean, only used in rc mode
122                'show_minor'   => array('bool', 'minor', false),
123                // String, only used in list mode
124                'sort'         => array('str', 'sort', 'natural'),
125                // String, only used in search mode
126                'search_query' => array('str', 'q', null),
127                // One of: pages, media, both
128                'content_type' => array('str', 'view', $conf['rss_media'])
129
130            ) as $name => $val) {
131        $opt[$name] = $INPUT->$val[0]($val[1], $val[2], true);
132    }
133
134    $opt['items']      = max(0, (int) $opt['items']);
135    $opt['show_minor'] = (bool) $opt['show_minor'];
136    $opt['sort'] = valid_input_set('sort', array('default' => 'natural', 'date'), $opt);
137
138    $opt['guardmail'] = ($conf['mailguard'] != '' && $conf['mailguard'] != 'none');
139
140    $type = valid_input_set(
141        'type', array(
142                     'rss', 'rss2', 'atom', 'atom1', 'rss1',
143                     'default' => $conf['rss_type']
144                ),
145        $_REQUEST
146    );
147    switch($type) {
148        case 'rss':
149            $opt['feed_type'] = 'RSS0.91';
150            $opt['mime_type'] = 'text/xml';
151            break;
152        case 'rss2':
153            $opt['feed_type'] = 'RSS2.0';
154            $opt['mime_type'] = 'text/xml';
155            break;
156        case 'atom':
157            $opt['feed_type'] = 'ATOM0.3';
158            $opt['mime_type'] = 'application/xml';
159            break;
160        case 'atom1':
161            $opt['feed_type'] = 'ATOM1.0';
162            $opt['mime_type'] = 'application/atom+xml';
163            break;
164        default:
165            $opt['feed_type'] = 'RSS1.0';
166            $opt['mime_type'] = 'application/xml';
167    }
168
169    $eventData = array(
170        'opt' => &$opt,
171    );
172    trigger_event('FEED_OPTS_POSTPROCESS', $eventData);
173    return $opt;
174}
175
176/**
177 * Add recent changed pages to a feed object
178 *
179 * @author Andreas Gohr <andi@splitbrain.org>
180 * @param  FeedCreator $rss the FeedCreator Object
181 * @param  array       $data the items to add
182 * @param  array       $opt  the feed options
183 */
184function rss_buildItems(&$rss, &$data, $opt) {
185    global $conf;
186    global $lang;
187    /* @var DokuWiki_Auth_Plugin $auth */
188    global $auth;
189
190    $eventData = array(
191        'rss'  => &$rss,
192        'data' => &$data,
193        'opt'  => &$opt,
194    );
195    $event     = new Doku_Event('FEED_DATA_PROCESS', $eventData);
196    if($event->advise_before(false)) {
197        foreach($data as $ditem) {
198            if(!is_array($ditem)) {
199                // not an array? then only a list of IDs was given
200                $ditem = array('id' => $ditem);
201            }
202
203            $item = new FeedItem();
204            $id   = $ditem['id'];
205            if(!$ditem['media']) {
206                $meta = p_get_metadata($id);
207            } else {
208                $meta = array();
209            }
210
211            // add date
212            if($ditem['date']) {
213                $date = $ditem['date'];
214            } elseif ($ditem['media']) {
215                $date = @filemtime(mediaFN($id));
216            } elseif (@file_exists(wikiFN($id))) {
217                $date = @filemtime(wikiFN($id));
218            } elseif($meta['date']['modified']) {
219                $date = $meta['date']['modified'];
220            } else {
221                $date = 0;
222            }
223            if($date) $item->date = date('r', $date);
224
225            // add title
226            if($conf['useheading'] && $meta['title']) {
227                $item->title = $meta['title'];
228            } else {
229                $item->title = $ditem['id'];
230            }
231            if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
232                $item->title .= ' - '.strip_tags($ditem['sum']);
233            }
234
235            // add item link
236            switch($opt['link_to']) {
237                case 'page':
238                    if($ditem['media']) {
239                        $item->link = media_managerURL(
240                            array(
241                                 'image' => $id,
242                                 'ns'    => getNS($id),
243                                 'rev'   => $date
244                            ), '&', true
245                        );
246                    } else {
247                        $item->link = wl($id, 'rev='.$date, true, '&');
248                    }
249                    break;
250                case 'rev':
251                    if($ditem['media']) {
252                        $item->link = media_managerURL(
253                            array(
254                                 'image'       => $id,
255                                 'ns'          => getNS($id),
256                                 'rev'         => $date,
257                                 'tab_details' => 'history'
258                            ), '&', true
259                        );
260                    } else {
261                        $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
262                    }
263                    break;
264                case 'current':
265                    if($ditem['media']) {
266                        $item->link = media_managerURL(
267                            array(
268                                 'image' => $id,
269                                 'ns'    => getNS($id)
270                            ), '&', true
271                        );
272                    } else {
273                        $item->link = wl($id, '', true, '&');
274                    }
275                    break;
276                case 'diff':
277                default:
278                    if($ditem['media']) {
279                        $item->link = media_managerURL(
280                            array(
281                                 'image'       => $id,
282                                 'ns'          => getNS($id),
283                                 'rev'         => $date,
284                                 'tab_details' => 'history',
285                                 'mediado'     => 'diff'
286                            ), '&', true
287                        );
288                    } else {
289                        $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
290                    }
291            }
292
293            // add item content
294            switch($opt['item_content']) {
295                case 'diff':
296                case 'htmldiff':
297                    if($ditem['media']) {
298                        $medialog = new MediaChangeLog($id);
299                        $revs  = $medialog->getRevisions(0, 1);
300                        $rev   = $revs[0];
301                        $src_r = '';
302                        $src_l = '';
303
304                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)), 300)) {
305                            $more  = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
306                            $src_r = ml($id, $more, true, '&amp;', true);
307                        }
308                        if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
309                            $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
310                            $src_l = ml($id, $more, true, '&amp;', true);
311                        }
312                        $content = '';
313                        if($src_r) {
314                            $content = '<table>';
315                            $content .= '<tr><th width="50%">'.$rev.'</th>';
316                            $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
317                            $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
318                            $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
319                            $content .= '</table>';
320                        }
321
322                    } else {
323                        require_once(DOKU_INC.'inc/DifferenceEngine.php');
324                        $pagelog = new PageChangeLog($id);
325                        $revs = $pagelog->getRevisions(0, 1);
326                        $rev  = $revs[0];
327
328                        if($rev) {
329                            $df = new Diff(explode("\n", rawWiki($id, $rev)),
330                                           explode("\n", rawWiki($id, '')));
331                        } else {
332                            $df = new Diff(array(''),
333                                           explode("\n", rawWiki($id, '')));
334                        }
335
336                        if($opt['item_content'] == 'htmldiff') {
337                            // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
338                            $tdf     = new TableDiffFormatter();
339                            $content = '<table>';
340                            $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
341                            $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
342                            $content .= $tdf->format($df);
343                            $content .= '</table>';
344                        } else {
345                            // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
346                            $udf     = new UnifiedDiffFormatter();
347                            $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
348                        }
349                    }
350                    break;
351                case 'html':
352                    if($ditem['media']) {
353                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
354                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
355                            $src     = ml($id, $more, true, '&amp;', true);
356                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
357                        } else {
358                            $content = '';
359                        }
360                    } else {
361                        if (@filemtime(wikiFN($id)) === $date) {
362                            $content = p_wiki_xhtml($id, '', false);
363                        } else {
364                            $content = p_wiki_xhtml($id, $date, false);
365                        }
366                        // no TOC in feeds
367                        $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
368
369                        // add alignment for images
370                        $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
371                        $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
372
373                        // make URLs work when canonical is not set, regexp instead of rerendering!
374                        if(!$conf['canonical']) {
375                            $base    = preg_quote(DOKU_REL, '/');
376                            $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
377                        }
378                    }
379
380                    break;
381                case 'abstract':
382                default:
383                    if($ditem['media']) {
384                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
385                            $more    = 'w='.$size[0].'&h='.$size[1].'&t='.@filemtime(mediaFN($id));
386                            $src     = ml($id, $more, true, '&amp;', true);
387                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
388                        } else {
389                            $content = '';
390                        }
391                    } else {
392                        $content = $meta['description']['abstract'];
393                    }
394            }
395            $item->description = $content; //FIXME a plugin hook here could be senseful
396
397            // add user
398            # FIXME should the user be pulled from metadata as well?
399            $user         = @$ditem['user']; // the @ spares time repeating lookup
400            $item->author = '';
401            if($user && $conf['useacl'] && $auth) {
402                $userInfo = $auth->getUserData($user);
403                if($userInfo) {
404                    switch($conf['showuseras']) {
405                        case 'username':
406                            $item->author = $userInfo['name'];
407                            break;
408                        default:
409                            $item->author = $user;
410                            break;
411                    }
412                } else {
413                    $item->author = $user;
414                }
415                if($userInfo && !$opt['guardmail']) {
416                    $item->authorEmail = $userInfo['mail'];
417                } else {
418                    //cannot obfuscate because some RSS readers may check validity
419                    $item->authorEmail = $user.'@'.$ditem['ip'];
420                }
421            } elseif($user) {
422                // this happens when no ACL but some Apache auth is used
423                $item->author      = $user;
424                $item->authorEmail = $user.'@'.$ditem['ip'];
425            } else {
426                $item->authorEmail = 'anonymous@'.$ditem['ip'];
427            }
428
429            // add category
430            if(isset($meta['subject'])) {
431                $item->category = $meta['subject'];
432            } else {
433                $cat = getNS($id);
434                if($cat) $item->category = $cat;
435            }
436
437            // finally add the item to the feed object, after handing it to registered plugins
438            $evdata = array(
439                'item'  => &$item,
440                'opt'   => &$opt,
441                'ditem' => &$ditem,
442                'rss'   => &$rss
443            );
444            $evt    = new Doku_Event('FEED_ITEM_ADD', $evdata);
445            if($evt->advise_before()) {
446                $rss->addItem($item);
447            }
448            $evt->advise_after(); // for completeness
449        }
450    }
451    $event->advise_after();
452}
453
454/**
455 * Add recent changed pages to the feed object
456 *
457 * @author Andreas Gohr <andi@splitbrain.org>
458 */
459function rssRecentChanges($opt) {
460    global $conf;
461    $flags = RECENTS_SKIP_DELETED;
462    if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
463    if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
464    if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
465
466    $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
467    return $recents;
468}
469
470/**
471 * Add all pages of a namespace to the feed object
472 *
473 * @author Andreas Gohr <andi@splitbrain.org>
474 */
475function rssListNamespace($opt) {
476    require_once(DOKU_INC.'inc/search.php');
477    global $conf;
478
479    $ns = ':'.cleanID($opt['namespace']);
480    $ns = utf8_encodeFN(str_replace(':', '/', $ns));
481
482    $data = array();
483    $search_opts = array(
484        'depth' => 1,
485        'pagesonly' => true,
486        'listfiles' => true
487    );
488    search($data, $conf['datadir'], 'search_universal', $search_opts, $ns, $lvl = 1, $opt['sort']);
489
490    return $data;
491}
492
493/**
494 * Add the result of a full text search to the feed object
495 *
496 * @author Andreas Gohr <andi@splitbrain.org>
497 */
498function rssSearch($opt) {
499    if(!$opt['search_query']) return array();
500
501    require_once(DOKU_INC.'inc/fulltext.php');
502    $data = ft_pageSearch($opt['search_query'], $poswords);
503    $data = array_keys($data);
504
505    return $data;
506}
507
508//Setup VIM: ex: et ts=4 :
509