xref: /dokuwiki/feed.php (revision 1306d063bc48af5df32e884a0a166dc34376b5b3)
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$cache = getCacheName(join('', array_values($opt)).$_SERVER['REMOTE_USER'], '.feed');
24$key   = join('', array_values($opt)).$_SERVER['REMOTE_USER'];
25$cache = new cache($key, '.feed');
26
27// prepare cache depends
28$depends['files'] = getConfigFiles('main');
29$depends['age']   = $conf['rss_update'];
30$depends['purge'] = $INPUT->bool('purge');
31
32// check cacheage and deliver if nothing has changed since last
33// time or the update interval has not passed, also handles conditional requests
34header('Cache-Control: must-revalidate, post-check=0, pre-check=0');
35header('Pragma: public');
36header('Content-Type: application/xml; charset=utf-8');
37header('X-Robots-Tag: noindex');
38if($cache->useCache($depends)) {
39    http_conditionalRequest($cache->_time);
40    if($conf['allowdebug']) header("X-CacheUsed: $cache->cache");
41    print $cache->retrieveCache();
42    exit;
43} else {
44    http_conditionalRequest(time());
45}
46
47// create new feed
48$rss                 = new DokuWikiFeedCreator();
49$rss->title          = $conf['title'].(($opt['namespace']) ? ' '.$opt['namespace'] : '');
50$rss->link           = DOKU_URL;
51$rss->syndicationURL = DOKU_URL.'feed.php';
52$rss->cssStyleSheet  = DOKU_URL.'lib/exe/css.php?s=feed';
53
54$image        = new FeedImage();
55$image->title = $conf['title'];
56$image->url   = tpl_getMediaFile(array(':wiki:favicon.ico', ':favicon.ico', 'images/favicon.ico'), true);
57$image->link  = DOKU_URL;
58$rss->image   = $image;
59
60$data  = null;
61$modes = array(
62    'list'   => 'rssListNamespace',
63    'search' => 'rssSearch',
64    'recent' => 'rssRecentChanges'
65);
66if(isset($modes[$opt['feed_mode']])) {
67    $data = $modes[$opt['feed_mode']]($opt);
68} else {
69    $eventData = array(
70        'opt'  => &$opt,
71        'data' => &$data,
72    );
73    $event     = new Doku_Event('FEED_MODE_UNKNOWN', $eventData);
74    if($event->advise_before(true)) {
75        echo sprintf('<error>Unknown feed mode %s</error>', hsc($opt['feed_mode']));
76        exit;
77    }
78    $event->advise_after();
79}
80
81rss_buildItems($rss, $data, $opt);
82$feed = $rss->createFeed($opt['feed_type'], 'utf-8');
83
84// save cachefile
85$cache->storeCache($feed);
86
87// finally deliver
88print $feed;
89
90// ---------------------------------------------------------------- //
91
92/**
93 * Get URL parameters and config options and return an initialized option array
94 *
95 * @author Andreas Gohr <andi@splitbrain.org>
96 */
97function rss_parseOptions() {
98    global $conf;
99    global $INPUT;
100
101    $opt = array();
102
103    foreach(array(
104                // Basic feed properties
105                // Plugins may probably want to add new values to these
106                // properties for implementing own feeds
107
108                // One of: list, search, recent
109                'feed_mode'    => array('str', 'mode', 'recent'),
110                // One of: diff, page, rev, current
111                'link_to'      => array('str', 'linkto', $conf['rss_linkto']),
112                // One of: abstract, diff, htmldiff, html
113                'item_content' => array('str', 'content', $conf['rss_content']),
114
115                // Special feed properties
116                // These are only used by certain feed_modes
117
118                // String, used for feed title, in list and rc mode
119                'namespace'    => array('str', 'ns', null),
120                // Positive integer, only used in rc mode
121                'items'        => array('int', 'num', $conf['recent']),
122                // Boolean, only used in rc mode
123                'show_minor'   => array('bool', 'minor', false),
124                // String, only used in search mode
125                'search_query' => array('str', 'q', null),
126                // One of: pages, media, both
127                'content_type' => array('str', 'view', $conf['rss_media'])
128
129            ) as $name => $val) {
130        $opt[$name] = $INPUT->$val[0]($val[1], $val[2], true);
131    }
132
133    $opt['items']      = max(0, (int) $opt['items']);
134    $opt['show_minor'] = (bool) $opt['show_minor'];
135
136    $opt['guardmail'] = ($conf['mailguard'] != '' && $conf['mailguard'] != 'none');
137
138    $type = valid_input_set(
139        'type', array(
140                     'rss', 'rss2', 'atom', 'atom1', 'rss1',
141                     'default' => $conf['rss_type']
142                ),
143        $_REQUEST
144    );
145    switch($type) {
146        case 'rss':
147            $opt['feed_type'] = 'RSS0.91';
148            $opt['mime_type'] = 'text/xml';
149            break;
150        case 'rss2':
151            $opt['feed_type'] = 'RSS2.0';
152            $opt['mime_type'] = 'text/xml';
153            break;
154        case 'atom':
155            $opt['feed_type'] = 'ATOM0.3';
156            $opt['mime_type'] = 'application/xml';
157            break;
158        case 'atom1':
159            $opt['feed_type'] = 'ATOM1.0';
160            $opt['mime_type'] = 'application/atom+xml';
161            break;
162        default:
163            $opt['feed_type'] = 'RSS1.0';
164            $opt['mime_type'] = 'application/xml';
165    }
166
167    $eventData = array(
168        'opt' => &$opt,
169    );
170    trigger_event('FEED_OPTS_POSTPROCESS', $eventData);
171    return $opt;
172}
173
174/**
175 * Add recent changed pages to a feed object
176 *
177 * @author Andreas Gohr <andi@splitbrain.org>
178 * @param  FeedCreator $rss the FeedCreator Object
179 * @param  array       $data the items to add
180 * @param  array       $opt  the feed options
181 */
182function rss_buildItems(&$rss, &$data, $opt) {
183    global $conf;
184    global $lang;
185    /* @var auth_basic $auth */
186    global $auth;
187
188    $eventData = array(
189        'rss'  => &$rss,
190        'data' => &$data,
191        'opt'  => &$opt,
192    );
193    $event     = new Doku_Event('FEED_DATA_PROCESS', $eventData);
194    if($event->advise_before(false)) {
195        foreach($data as $ditem) {
196            if(!is_array($ditem)) {
197                // not an array? then only a list of IDs was given
198                $ditem = array('id' => $ditem);
199            }
200
201            $item = new FeedItem();
202            $id   = $ditem['id'];
203            if(!$ditem['media']) {
204                $meta = p_get_metadata($id);
205            } else {
206                $meta = array();
207            }
208
209            // add date
210            if($ditem['date']) {
211                $date = $ditem['date'];
212            } elseif ($ditem['media']) {
213                $date = @filemtime(mediaFN($id));
214            } elseif (@file_exists(wikiFN($id))) {
215                $date = @filemtime(wikiFN($id));
216            } elseif($meta['date']['modified']) {
217                $date = $meta['date']['modified'];
218            } else {
219                $date = 0;
220            }
221            if($date) $item->date = date('r', $date);
222
223            // add title
224            if($conf['useheading'] && $meta['title']) {
225                $item->title = $meta['title'];
226            } else {
227                $item->title = $ditem['id'];
228            }
229            if($conf['rss_show_summary'] && !empty($ditem['sum'])) {
230                $item->title .= ' - '.strip_tags($ditem['sum']);
231            }
232
233            // add item link
234            switch($opt['link_to']) {
235                case 'page':
236                    if($ditem['media']) {
237                        $item->link = media_managerURL(
238                            array(
239                                 'image' => $id,
240                                 'ns'    => getNS($id),
241                                 'rev'   => $date
242                            ), '&', true
243                        );
244                    } else {
245                        $item->link = wl($id, 'rev='.$date, true, '&');
246                    }
247                    break;
248                case 'rev':
249                    if($ditem['media']) {
250                        $item->link = media_managerURL(
251                            array(
252                                 'image'       => $id,
253                                 'ns'          => getNS($id),
254                                 'rev'         => $date,
255                                 'tab_details' => 'history'
256                            ), '&', true
257                        );
258                    } else {
259                        $item->link = wl($id, 'do=revisions&rev='.$date, true, '&');
260                    }
261                    break;
262                case 'current':
263                    if($ditem['media']) {
264                        $item->link = media_managerURL(
265                            array(
266                                 'image' => $id,
267                                 'ns'    => getNS($id)
268                            ), '&', true
269                        );
270                    } else {
271                        $item->link = wl($id, '', true, '&');
272                    }
273                    break;
274                case 'diff':
275                default:
276                    if($ditem['media']) {
277                        $item->link = media_managerURL(
278                            array(
279                                 'image'       => $id,
280                                 'ns'          => getNS($id),
281                                 'rev'         => $date,
282                                 'tab_details' => 'history',
283                                 'mediado'     => 'diff'
284                            ), '&', true
285                        );
286                    } else {
287                        $item->link = wl($id, 'rev='.$date.'&do=diff', true, '&');
288                    }
289            }
290
291            // add item content
292            switch($opt['item_content']) {
293                case 'diff':
294                case 'htmldiff':
295                    if($ditem['media']) {
296                        $revs  = getRevisions($id, 0, 1, 8192, true);
297                        $rev   = $revs[0];
298                        $src_r = '';
299                        $src_l = '';
300
301                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)), 300)) {
302                            $more  = 'w='.$size[0].'&h='.$size[1].'t='.@filemtime(mediaFN($id));
303                            $src_r = ml($id, $more);
304                        }
305                        if($rev && $size = media_image_preview_size($id, $rev, new JpegMeta(mediaFN($id, $rev)), 300)) {
306                            $more  = 'rev='.$rev.'&w='.$size[0].'&h='.$size[1];
307                            $src_l = ml($id, $more);
308                        }
309                        $content = '';
310                        if($src_r) {
311                            $content = '<table>';
312                            $content .= '<tr><th width="50%">'.$rev.'</th>';
313                            $content .= '<th width="50%">'.$lang['current'].'</th></tr>';
314                            $content .= '<tr align="center"><td><img src="'.$src_l.'" alt="" /></td><td>';
315                            $content .= '<img src="'.$src_r.'" alt="'.$id.'" /></td></tr>';
316                            $content .= '</table>';
317                        }
318
319                    } else {
320                        require_once(DOKU_INC.'inc/DifferenceEngine.php');
321                        $revs = getRevisions($id, 0, 1);
322                        $rev  = $revs[0];
323
324                        if($rev) {
325                            $df = new Diff(explode("\n", rawWiki($id, $rev)),
326                                           explode("\n", rawWiki($id, '')));
327                        } else {
328                            $df = new Diff(array(''),
329                                           explode("\n", rawWiki($id, '')));
330                        }
331
332                        if($opt['item_content'] == 'htmldiff') {
333                            // note: no need to escape diff output, TableDiffFormatter provides 'safe' html
334                            $tdf     = new TableDiffFormatter();
335                            $content = '<table>';
336                            $content .= '<tr><th colspan="2" width="50%">'.$rev.'</th>';
337                            $content .= '<th colspan="2" width="50%">'.$lang['current'].'</th></tr>';
338                            $content .= $tdf->format($df);
339                            $content .= '</table>';
340                        } else {
341                            // note: diff output must be escaped, UnifiedDiffFormatter provides plain text
342                            $udf     = new UnifiedDiffFormatter();
343                            $content = "<pre>\n".hsc($udf->format($df))."\n</pre>";
344                        }
345                    }
346                    break;
347                case 'html':
348                    if($ditem['media']) {
349                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
350                            $more    = 'w='.$size[0].'&h='.$size[1].'t='.@filemtime(mediaFN($id));
351                            $src     = ml($id, $more);
352                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
353                        } else {
354                            $content = '';
355                        }
356                    } else {
357                        if (@filemtime(wikiFN($id)) === $date) {
358                            $content = p_wiki_xhtml($id, '', false);
359                        } else {
360                            $content = p_wiki_xhtml($id, $date, false);
361                        }
362                        // no TOC in feeds
363                        $content = preg_replace('/(<!-- TOC START -->).*(<!-- TOC END -->)/s', '', $content);
364
365                        // add alignment for images
366                        $content = preg_replace('/(<img .*?class="medialeft")/s', '\\1 align="left"', $content);
367                        $content = preg_replace('/(<img .*?class="mediaright")/s', '\\1 align="right"', $content);
368
369                        // make URLs work when canonical is not set, regexp instead of rerendering!
370                        if(!$conf['canonical']) {
371                            $base    = preg_quote(DOKU_REL, '/');
372                            $content = preg_replace('/(<a href|<img src)="('.$base.')/s', '$1="'.DOKU_URL, $content);
373                        }
374                    }
375
376                    break;
377                case 'abstract':
378                default:
379                    if($ditem['media']) {
380                        if($size = media_image_preview_size($id, false, new JpegMeta(mediaFN($id)))) {
381                            $more    = 'w='.$size[0].'&h='.$size[1].'t='.@filemtime(mediaFN($id));
382                            $src     = ml($id, $more);
383                            $content = '<img src="'.$src.'" alt="'.$id.'" />';
384                        } else {
385                            $content = '';
386                        }
387                    } else {
388                        $content = $meta['description']['abstract'];
389                    }
390            }
391            $item->description = $content; //FIXME a plugin hook here could be senseful
392
393            // add user
394            # FIXME should the user be pulled from metadata as well?
395            $user         = @$ditem['user']; // the @ spares time repeating lookup
396            $item->author = '';
397            if($user && $conf['useacl'] && $auth) {
398                $userInfo = $auth->getUserData($user);
399                if($userInfo) {
400                    switch($conf['showuseras']) {
401                        case 'username':
402                            $item->author = $userInfo['name'];
403                            break;
404                        default:
405                            $item->author = $user;
406                            break;
407                    }
408                } else {
409                    $item->author = $user;
410                }
411                if($userInfo && !$opt['guardmail']) {
412                    $item->authorEmail = $userInfo['mail'];
413                } else {
414                    //cannot obfuscate because some RSS readers may check validity
415                    $item->authorEmail = $user.'@'.$ditem['ip'];
416                }
417            } elseif($user) {
418                // this happens when no ACL but some Apache auth is used
419                $item->author      = $user;
420                $item->authorEmail = $user.'@'.$ditem['ip'];
421            } else {
422                $item->authorEmail = 'anonymous@'.$ditem['ip'];
423            }
424
425            // add category
426            if(isset($meta['subject'])) {
427                $item->category = $meta['subject'];
428            } else {
429                $cat = getNS($id);
430                if($cat) $item->category = $cat;
431            }
432
433            // finally add the item to the feed object, after handing it to registered plugins
434            $evdata = array(
435                'item'  => &$item,
436                'opt'   => &$opt,
437                'ditem' => &$ditem,
438                'rss'   => &$rss
439            );
440            $evt    = new Doku_Event('FEED_ITEM_ADD', $evdata);
441            if($evt->advise_before()) {
442                $rss->addItem($item);
443            }
444            $evt->advise_after(); // for completeness
445        }
446    }
447    $event->advise_after();
448}
449
450/**
451 * Add recent changed pages to the feed object
452 *
453 * @author Andreas Gohr <andi@splitbrain.org>
454 */
455function rssRecentChanges($opt) {
456    global $conf;
457    $flags = RECENTS_SKIP_DELETED;
458    if(!$opt['show_minor']) $flags += RECENTS_SKIP_MINORS;
459    if($opt['content_type'] == 'media' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_CHANGES;
460    if($opt['content_type'] == 'both' && $conf['mediarevisions']) $flags += RECENTS_MEDIA_PAGES_MIXED;
461
462    $recents = getRecents(0, $opt['items'], $opt['namespace'], $flags);
463    return $recents;
464}
465
466/**
467 * Add all pages of a namespace to the feed object
468 *
469 * @author Andreas Gohr <andi@splitbrain.org>
470 */
471function rssListNamespace($opt) {
472    require_once(DOKU_INC.'inc/search.php');
473    global $conf;
474
475    $ns = ':'.cleanID($opt['namespace']);
476    $ns = str_replace(':', '/', $ns);
477
478    $data = array();
479    sort($data);
480    search($data, $conf['datadir'], 'search_list', '', $ns);
481
482    return $data;
483}
484
485/**
486 * Add the result of a full text search to the feed object
487 *
488 * @author Andreas Gohr <andi@splitbrain.org>
489 */
490function rssSearch($opt) {
491    if(!$opt['search_query']) return array();
492
493    require_once(DOKU_INC.'inc/fulltext.php');
494    $data = ft_pageSearch($opt['search_query'], $poswords);
495    $data = array_keys($data);
496
497    return $data;
498}
499
500//Setup VIM: ex: et ts=4 :
501