xref: /dokuwiki/lib/exe/indexer.php (revision 38dc5fa6df9a0cd5afd16dc95600d7cf11b08877) 
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// Version tag used to force rebuild on upgrade
15define('INDEXER_VERSION', 2);
16
17// keep running after browser closes connection
18@ignore_user_abort(true);
19
20// check if user abort worked, if yes send output early
21$defer = !@ignore_user_abort() || $conf['broken_iua'];
22if(!$defer){
23    sendGIF(); // send gif
24}
25
26$ID = cleanID($_REQUEST['id']);
27
28// Catch any possible output (e.g. errors)
29$output = isset($_REQUEST['debug']) && $conf['allowdebug'];
30if(!$output) ob_start();
31
32// run one of the jobs
33$tmp = array(); // No event data
34$evt = new Doku_Event('INDEXER_TASKS_RUN', $tmp);
35if ($evt->advise_before()) {
36  runIndexer() or
37  runSitemapper() or
38  sendDigest() or
39  runTrimRecentChanges() or
40  runTrimRecentChanges(true) or
41  $evt->advise_after();
42}
43if($defer) sendGIF();
44
45if(!$output) ob_end_clean();
46exit;
47
48// --------------------------------------------------------------------
49
50/**
51 * Trims the recent changes cache (or imports the old changelog) as needed.
52 *
53 * @param media_changes If the media changelog shall be trimmed instead of
54 * the page changelog
55 *
56 * @author Ben Coburn <btcoburn@silicodon.net>
57 */
58function runTrimRecentChanges($media_changes = false) {
59    global $conf;
60
61    $fn = ($media_changes ? $conf['media_changelog'] : $conf['changelog']);
62
63    // Trim the Recent Changes
64    // Trims the recent changes cache to the last $conf['changes_days'] recent
65    // changes or $conf['recent'] items, which ever is larger.
66    // The trimming is only done once a day.
67    if (@file_exists($fn) &&
68        (@filemtime($fn.'.trimmed')+86400)<time() &&
69        !@file_exists($fn.'_tmp')) {
70            @touch($fn.'.trimmed');
71            io_lock($fn);
72            $lines = file($fn);
73            if (count($lines)<=$conf['recent']) {
74                // nothing to trim
75                io_unlock($fn);
76                return false;
77            }
78
79            io_saveFile($fn.'_tmp', '');          // presave tmp as 2nd lock
80            $trim_time = time() - $conf['recent_days']*86400;
81            $out_lines = array();
82
83            for ($i=0; $i<count($lines); $i++) {
84              $log = parseChangelogLine($lines[$i]);
85              if ($log === false) continue;                      // discard junk
86              if ($log['date'] < $trim_time) {
87                $old_lines[$log['date'].".$i"] = $lines[$i];     // keep old lines for now (append .$i to prevent key collisions)
88              } else {
89                $out_lines[$log['date'].".$i"] = $lines[$i];     // definitely keep these lines
90              }
91            }
92
93            if (count($lines)==count($out_lines)) {
94              // nothing to trim
95              @unlink($fn.'_tmp');
96              io_unlock($fn);
97              return false;
98            }
99
100            // sort the final result, it shouldn't be necessary,
101            //   however the extra robustness in making the changelog cache self-correcting is worth it
102            ksort($out_lines);
103            $extra = $conf['recent'] - count($out_lines);        // do we need extra lines do bring us up to minimum
104            if ($extra > 0) {
105              ksort($old_lines);
106              $out_lines = array_merge(array_slice($old_lines,-$extra),$out_lines);
107            }
108
109            // save trimmed changelog
110            io_saveFile($fn.'_tmp', implode('', $out_lines));
111            @unlink($fn);
112            if (!rename($fn.'_tmp', $fn)) {
113                // rename failed so try another way...
114                io_unlock($fn);
115                io_saveFile($fn, implode('', $out_lines));
116                @unlink($fn.'_tmp');
117            } else {
118                io_unlock($fn);
119            }
120            return true;
121    }
122
123    // nothing done
124    return false;
125}
126
127/**
128 * Runs the indexer for the current page
129 *
130 * @author Andreas Gohr <andi@splitbrain.org>
131 */
132function runIndexer(){
133    global $ID;
134    global $conf;
135    print "runIndexer(): started".NL;
136
137    if(!$ID) return false;
138
139    // check if indexing needed
140    $idxtag = metaFN($ID,'.indexed');
141    if(@file_exists($idxtag)){
142        if(io_readFile($idxtag) >= INDEXER_VERSION){
143            $last = @filemtime($idxtag);
144            if($last > @filemtime(wikiFN($ID))){
145                print "runIndexer(): index for $ID up to date".NL;
146                return false;
147            }
148        }
149    }
150
151    // try to aquire a lock
152    $lock = $conf['lockdir'].'/_indexer.lock';
153    while(!@mkdir($lock,$conf['dmode'])){
154        usleep(50);
155        if(time()-@filemtime($lock) > 60*5){
156            // looks like a stale lock - remove it
157            @rmdir($lock);
158            print "runIndexer(): stale lock removed".NL;
159        }else{
160            print "runIndexer(): indexer locked".NL;
161            return false;
162        }
163    }
164    if($conf['dperm']) chmod($lock, $conf['dperm']);
165
166    // do the work
167    idx_addPage($ID);
168
169    // we're finished - save and free lock
170    io_saveFile(metaFN($ID,'.indexed'),INDEXER_VERSION);
171    @rmdir($lock);
172    print "runIndexer(): finished".NL;
173    return true;
174}
175
176/**
177 * Builds a Google Sitemap of all public pages known to the indexer
178 *
179 * The map is placed in the root directory named sitemap.xml.gz - This
180 * file needs to be writable!
181 *
182 * @author Andreas Gohr
183 * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
184 */
185function runSitemapper(){
186    print "runSitemapper(): started".NL;
187    $result = Sitemapper::generate() && Sitemapper::pingSearchEngines();
188    print 'runSitemapper(): finished'.NL;
189    return $result;
190}
191
192/**
193 * Send digest and list mails for all subscriptions which are in effect for the
194 * current page
195 *
196 * @author Adrian Lang <lang@cosmocode.de>
197 */
198function sendDigest() {
199    echo 'sendDigest(): start'.NL;
200    global $ID;
201    global $conf;
202    if (!$conf['subscribers']) {
203        return;
204    }
205    $subscriptions = subscription_find($ID, array('style' => '(digest|list)',
206                                                  'escaped' => true));
207    global $auth;
208    global $lang;
209    global $conf;
210    global $USERINFO;
211
212    // remember current user info
213    $olduinfo = $USERINFO;
214    $olduser  = $_SERVER['REMOTE_USER'];
215
216    foreach($subscriptions as $id => $users) {
217        if (!subscription_lock($id)) {
218            continue;
219        }
220        foreach($users as $data) {
221            list($user, $style, $lastupdate) = $data;
222            $lastupdate = (int) $lastupdate;
223            if ($lastupdate + $conf['subscribe_time'] > time()) {
224                // Less than the configured time period passed since last
225                // update.
226                continue;
227            }
228
229            // Work as the user to make sure ACLs apply correctly
230            $USERINFO = $auth->getUserData($user);
231            $_SERVER['REMOTE_USER'] = $user;
232            if ($USERINFO === false) {
233                continue;
234            }
235
236            if (substr($id, -1, 1) === ':') {
237                // The subscription target is a namespace
238                $changes = getRecentsSince($lastupdate, null, getNS($id));
239            } else {
240                if(auth_quickaclcheck($id) < AUTH_READ) continue;
241
242                $meta = p_get_metadata($id);
243                $changes = array($meta['last_change']);
244            }
245
246            // Filter out pages only changed in small and own edits
247            $change_ids = array();
248            foreach($changes as $rev) {
249                $n = 0;
250                while (!is_null($rev) && $rev['date'] >= $lastupdate &&
251                       ($_SERVER['REMOTE_USER'] === $rev['user'] ||
252                        $rev['type'] === DOKU_CHANGE_TYPE_MINOR_EDIT)) {
253                    $rev = getRevisions($rev['id'], $n++, 1);
254                    $rev = (count($rev) > 0) ? $rev[0] : null;
255                }
256
257                if (!is_null($rev) && $rev['date'] >= $lastupdate) {
258                    // Some change was not a minor one and not by myself
259                    $change_ids[] = $rev['id'];
260                }
261            }
262
263            if ($style === 'digest') {
264                foreach($change_ids as $change_id) {
265                    subscription_send_digest($USERINFO['mail'], $change_id,
266                                             $lastupdate);
267                }
268            } elseif ($style === 'list') {
269                subscription_send_list($USERINFO['mail'], $change_ids, $id);
270            }
271            // TODO: Handle duplicate subscriptions.
272
273            // Update notification time.
274            subscription_set($user, $id, $style, time(), true);
275        }
276        subscription_unlock($id);
277    }
278
279    // restore current user info
280    $USERINFO = $olduinfo;
281    $_SERVER['REMOTE_USER'] = $olduser;
282}
283
284/**
285 * Just send a 1x1 pixel blank gif to the browser
286 *
287 * @author Andreas Gohr <andi@splitbrain.org>
288 * @author Harry Fuecks <fuecks@gmail.com>
289 */
290function sendGIF(){
291    if(isset($_REQUEST['debug'])){
292        header('Content-Type: text/plain');
293        return;
294    }
295    $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7');
296    header('Content-Type: image/gif');
297    header('Content-Length: '.strlen($img));
298    header('Connection: Close');
299    print $img;
300    flush();
301    // Browser should drop connection after this
302    // Thinks it's got the whole image
303}
304
305//Setup VIM: ex: et ts=4 :
306// No trailing PHP closing tag - no output please!
307// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php
308