xref: /dokuwiki/lib/exe/indexer.php (revision 7eaa7703a43b9f22f12be04e8580a3f2515fb146) 
1<?php
2/**
3 * DokuWiki indexer
4 *
5 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6 * @author     Andreas Gohr <andi@splitbrain.org>
7 */
8if(!defined('DOKU_INC')) define('DOKU_INC',dirname(__FILE__).'/../../');
9define('DOKU_DISABLE_GZIP_OUTPUT',1);
10require_once(DOKU_INC.'inc/init.php');
11session_write_close();  //close session
12if(!defined('NL')) define('NL',"\n");
13
14// keep running after browser closes connection
15@ignore_user_abort(true);
16
17// check if user abort worked, if yes send output early
18$defer = !@ignore_user_abort() || $conf['broken_iua'];
19if(!$defer){
20    sendGIF(); // send gif
21}
22
23$ID = cleanID($INPUT->str('id'));
24
25// Catch any possible output (e.g. errors)
26$output = $INPUT->has('debug') && $conf['allowdebug'];
27if(!$output) ob_start();
28
29// run one of the jobs
30$tmp = array(); // No event data
31$evt = new Doku_Event('INDEXER_TASKS_RUN', $tmp);
32if ($evt->advise_before()) {
33  runIndexer() or
34  runSitemapper() or
35  sendDigest() or
36  runTrimRecentChanges() or
37  runTrimRecentChanges(true) or
38  $evt->advise_after();
39}
40
41if(!$output) {
42    ob_end_clean();
43    if($defer) sendGIF();
44}
45
46exit;
47
48// --------------------------------------------------------------------
49
50/**
51 * Trims the recent changes cache (or imports the old changelog) as needed.
52 *
53 * @param media_changes If the media changelog shall be trimmed instead of
54 * the page changelog
55 *
56 * @author Ben Coburn <btcoburn@silicodon.net>
57 */
58function runTrimRecentChanges($media_changes = false) {
59    global $conf;
60
61    echo "runTrimRecentChanges($media_changes): started".NL;
62
63    $fn = ($media_changes ? $conf['media_changelog'] : $conf['changelog']);
64
65    // Trim the Recent Changes
66    // Trims the recent changes cache to the last $conf['changes_days'] recent
67    // changes or $conf['recent'] items, which ever is larger.
68    // The trimming is only done once a day.
69    if (@file_exists($fn) &&
70        (@filemtime($fn.'.trimmed')+86400)<time() &&
71        !@file_exists($fn.'_tmp')) {
72            @touch($fn.'.trimmed');
73            io_lock($fn);
74            $lines = file($fn);
75            if (count($lines)<=$conf['recent']) {
76                // nothing to trim
77                io_unlock($fn);
78                echo "runTrimRecentChanges($media_changes): finished".NL;
79                return false;
80            }
81
82            io_saveFile($fn.'_tmp', '');          // presave tmp as 2nd lock
83            $trim_time = time() - $conf['recent_days']*86400;
84            $out_lines = array();
85
86            for ($i=0; $i<count($lines); $i++) {
87              $log = parseChangelogLine($lines[$i]);
88              if ($log === false) continue;                      // discard junk
89              if ($log['date'] < $trim_time) {
90                $old_lines[$log['date'].".$i"] = $lines[$i];     // keep old lines for now (append .$i to prevent key collisions)
91              } else {
92                $out_lines[$log['date'].".$i"] = $lines[$i];     // definitely keep these lines
93              }
94            }
95
96            if (count($lines)==count($out_lines)) {
97              // nothing to trim
98              @unlink($fn.'_tmp');
99              io_unlock($fn);
100              echo "runTrimRecentChanges($media_changes): finished".NL;
101              return false;
102            }
103
104            // sort the final result, it shouldn't be necessary,
105            //   however the extra robustness in making the changelog cache self-correcting is worth it
106            ksort($out_lines);
107            $extra = $conf['recent'] - count($out_lines);        // do we need extra lines do bring us up to minimum
108            if ($extra > 0) {
109              ksort($old_lines);
110              $out_lines = array_merge(array_slice($old_lines,-$extra),$out_lines);
111            }
112
113            // save trimmed changelog
114            io_saveFile($fn.'_tmp', implode('', $out_lines));
115            @unlink($fn);
116            if (!rename($fn.'_tmp', $fn)) {
117                // rename failed so try another way...
118                io_unlock($fn);
119                io_saveFile($fn, implode('', $out_lines));
120                @unlink($fn.'_tmp');
121            } else {
122                io_unlock($fn);
123            }
124            echo "runTrimRecentChanges($media_changes): finished".NL;
125            return true;
126    }
127
128    // nothing done
129    echo "runTrimRecentChanges($media_changes): finished".NL;
130    return false;
131}
132
133/**
134 * Runs the indexer for the current page
135 *
136 * @author Andreas Gohr <andi@splitbrain.org>
137 */
138function runIndexer(){
139    global $ID;
140    global $conf;
141    print "runIndexer(): started".NL;
142
143    if(!$ID) return false;
144
145    // do the work
146    return idx_addPage($ID, true);
147}
148
149/**
150 * Builds a Google Sitemap of all public pages known to the indexer
151 *
152 * The map is placed in the root directory named sitemap.xml.gz - This
153 * file needs to be writable!
154 *
155 * @author Andreas Gohr
156 * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
157 */
158function runSitemapper(){
159    print "runSitemapper(): started".NL;
160    $result = Sitemapper::generate() && Sitemapper::pingSearchEngines();
161    print 'runSitemapper(): finished'.NL;
162    return $result;
163}
164
165/**
166 * Send digest and list mails for all subscriptions which are in effect for the
167 * current page
168 *
169 * @author Adrian Lang <lang@cosmocode.de>
170 */
171function sendDigest() {
172    echo 'sendDigest(): started'.NL;
173    global $ID;
174    global $conf;
175    if (!$conf['subscribers']) {
176        echo 'sendDigest(): disabled'.NL;
177        return false;
178    }
179    $subscriptions = subscription_find($ID, array('style' => '(digest|list)',
180                                                  'escaped' => true));
181    /** @var auth_basic $auth */
182    global $auth;
183    global $lang;
184    global $conf;
185    global $USERINFO;
186
187    $sent = false;
188
189    // remember current user info
190    $olduinfo = $USERINFO;
191    $olduser  = $_SERVER['REMOTE_USER'];
192
193    foreach($subscriptions as $id => $users) {
194        if (!subscription_lock($id)) {
195            continue;
196        }
197        foreach($users as $data) {
198            list($user, $style, $lastupdate) = $data;
199            $lastupdate = (int) $lastupdate;
200            if ($lastupdate + $conf['subscribe_time'] > time()) {
201                // Less than the configured time period passed since last
202                // update.
203                continue;
204            }
205
206            // Work as the user to make sure ACLs apply correctly
207            $USERINFO = $auth->getUserData($user);
208            $_SERVER['REMOTE_USER'] = $user;
209            if ($USERINFO === false) {
210                continue;
211            }
212
213            if (substr($id, -1, 1) === ':') {
214                // The subscription target is a namespace
215                $changes = getRecentsSince($lastupdate, null, getNS($id));
216            } else {
217                if(auth_quickaclcheck($id) < AUTH_READ) continue;
218
219                $meta = p_get_metadata($id);
220                $changes = array($meta['last_change']);
221            }
222
223            // Filter out pages only changed in small and own edits
224            $change_ids = array();
225            foreach($changes as $rev) {
226                $n = 0;
227                while (!is_null($rev) && $rev['date'] >= $lastupdate &&
228                       ($_SERVER['REMOTE_USER'] === $rev['user'] ||
229                        $rev['type'] === DOKU_CHANGE_TYPE_MINOR_EDIT)) {
230                    $rev = getRevisions($rev['id'], $n++, 1);
231                    $rev = (count($rev) > 0) ? $rev[0] : null;
232                }
233
234                if (!is_null($rev) && $rev['date'] >= $lastupdate) {
235                    // Some change was not a minor one and not by myself
236                    $change_ids[] = $rev['id'];
237                }
238            }
239
240            if ($style === 'digest') {
241                foreach($change_ids as $change_id) {
242                    subscription_send_digest($USERINFO['mail'], $change_id,
243                                             $lastupdate);
244                    $sent = true;
245                }
246            } elseif ($style === 'list') {
247                subscription_send_list($USERINFO['mail'], $change_ids, $id);
248                $sent = true;
249            }
250            // TODO: Handle duplicate subscriptions.
251
252            // Update notification time.
253            subscription_set($user, $id, $style, time(), true);
254        }
255        subscription_unlock($id);
256    }
257
258    // restore current user info
259    $USERINFO = $olduinfo;
260    $_SERVER['REMOTE_USER'] = $olduser;
261    echo 'sendDigest(): finished'.NL;
262    return $sent;
263}
264
265/**
266 * Just send a 1x1 pixel blank gif to the browser
267 *
268 * @author Andreas Gohr <andi@splitbrain.org>
269 * @author Harry Fuecks <fuecks@gmail.com>
270 */
271function sendGIF(){
272    global $INPUT;
273    if($INPUT->has('debug')){
274        header('Content-Type: text/plain');
275        return;
276    }
277    $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7');
278    header('Content-Type: image/gif');
279    header('Content-Length: '.strlen($img));
280    header('Connection: Close');
281    print $img;
282    flush();
283    // Browser should drop connection after this
284    // Thinks it's got the whole image
285}
286
287//Setup VIM: ex: et ts=4 :
288// No trailing PHP closing tag - no output please!
289// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php
290