source: documentation/trunk/packages/dokuwiki-2011-05-25a/lib/exe/indexer.php@ 25027

Last change on this file since 25027 was 25027, checked in by jmt12, 12 years ago

Adding the packages directory, and within it a configured version of dokuwiki all ready to run
File size: 8.5 KB
Line 
1<?php
2/**
3 * DokuWiki indexer
4 *
5 * @license GPL 2 (http://www.gnu.org/licenses/gpl.html)
6 * @author Andreas Gohr <[email protected]>
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$ID = cleanID($_REQUEST['id']);
24
25// Catch any possible output (e.g. errors)
26$output = isset($_REQUEST['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 $evt->advise_after();
39}
40if($defer) sendGIF();
41
42if(!$output) ob_end_clean();
43exit;
44
45// --------------------------------------------------------------------
46
47/**
48 * Trims the recent changes cache (or imports the old changelog) as needed.
49 *
50 * @param media_changes If the media changelog shall be trimmed instead of
51 * the page changelog
52 *
53 * @author Ben Coburn <[email protected]>
54 */
55function runTrimRecentChanges($media_changes = false) {
56 global $conf;
57
58 $fn = ($media_changes ? $conf['media_changelog'] : $conf['changelog']);
59
60 // Trim the Recent Changes
61 // Trims the recent changes cache to the last $conf['changes_days'] recent
62 // changes or $conf['recent'] items, which ever is larger.
63 // The trimming is only done once a day.
64 if (@file_exists($fn) &&
65 (@filemtime($fn.'.trimmed')+86400)<time() &&
66 !@file_exists($fn.'_tmp')) {
67 @touch($fn.'.trimmed');
68 io_lock($fn);
69 $lines = file($fn);
70 if (count($lines)<=$conf['recent']) {
71 // nothing to trim
72 io_unlock($fn);
73 return false;
74 }
75
76 io_saveFile($fn.'_tmp', ''); // presave tmp as 2nd lock
77 $trim_time = time() - $conf['recent_days']*86400;
78 $out_lines = array();
79
80 for ($i=0; $i<count($lines); $i++) {
81 $log = parseChangelogLine($lines[$i]);
82 if ($log === false) continue; // discard junk
83 if ($log['date'] < $trim_time) {
84 $old_lines[$log['date'].".$i"] = $lines[$i]; // keep old lines for now (append .$i to prevent key collisions)
85 } else {
86 $out_lines[$log['date'].".$i"] = $lines[$i]; // definitely keep these lines
87 }
88 }
89
90 if (count($lines)==count($out_lines)) {
91 // nothing to trim
92 @unlink($fn.'_tmp');
93 io_unlock($fn);
94 return false;
95 }
96
97 // sort the final result, it shouldn't be necessary,
98 // however the extra robustness in making the changelog cache self-correcting is worth it
99 ksort($out_lines);
100 $extra = $conf['recent'] - count($out_lines); // do we need extra lines do bring us up to minimum
101 if ($extra > 0) {
102 ksort($old_lines);
103 $out_lines = array_merge(array_slice($old_lines,-$extra),$out_lines);
104 }
105
106 // save trimmed changelog
107 io_saveFile($fn.'_tmp', implode('', $out_lines));
108 @unlink($fn);
109 if (!rename($fn.'_tmp', $fn)) {
110 // rename failed so try another way...
111 io_unlock($fn);
112 io_saveFile($fn, implode('', $out_lines));
113 @unlink($fn.'_tmp');
114 } else {
115 io_unlock($fn);
116 }
117 return true;
118 }
119
120 // nothing done
121 return false;
122}
123
124/**
125 * Runs the indexer for the current page
126 *
127 * @author Andreas Gohr <[email protected]>
128 */
129function runIndexer(){
130 global $ID;
131 global $conf;
132 print "runIndexer(): started".NL;
133
134 if(!$ID) return false;
135
136 // do the work
137 return idx_addPage($ID, true);
138}
139
140/**
141 * Builds a Google Sitemap of all public pages known to the indexer
142 *
143 * The map is placed in the root directory named sitemap.xml.gz - This
144 * file needs to be writable!
145 *
146 * @author Andreas Gohr
147 * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html
148 */
149function runSitemapper(){
150 print "runSitemapper(): started".NL;
151 $result = Sitemapper::generate() && Sitemapper::pingSearchEngines();
152 print 'runSitemapper(): finished'.NL;
153 return $result;
154}
155
156/**
157 * Send digest and list mails for all subscriptions which are in effect for the
158 * current page
159 *
160 * @author Adrian Lang <[email protected]>
161 */
162function sendDigest() {
163 echo 'sendDigest(): start'.NL;
164 global $ID;
165 global $conf;
166 if (!$conf['subscribers']) {
167 return;
168 }
169 $subscriptions = subscription_find($ID, array('style' => '(digest|list)',
170 'escaped' => true));
171 global $auth;
172 global $lang;
173 global $conf;
174 global $USERINFO;
175
176 // remember current user info
177 $olduinfo = $USERINFO;
178 $olduser = $_SERVER['REMOTE_USER'];
179
180 foreach($subscriptions as $id => $users) {
181 if (!subscription_lock($id)) {
182 continue;
183 }
184 foreach($users as $data) {
185 list($user, $style, $lastupdate) = $data;
186 $lastupdate = (int) $lastupdate;
187 if ($lastupdate + $conf['subscribe_time'] > time()) {
188 // Less than the configured time period passed since last
189 // update.
190 continue;
191 }
192
193 // Work as the user to make sure ACLs apply correctly
194 $USERINFO = $auth->getUserData($user);
195 $_SERVER['REMOTE_USER'] = $user;
196 if ($USERINFO === false) {
197 continue;
198 }
199
200 if (substr($id, -1, 1) === ':') {
201 // The subscription target is a namespace
202 $changes = getRecentsSince($lastupdate, null, getNS($id));
203 } else {
204 if(auth_quickaclcheck($id) < AUTH_READ) continue;
205
206 $meta = p_get_metadata($id);
207 $changes = array($meta['last_change']);
208 }
209
210 // Filter out pages only changed in small and own edits
211 $change_ids = array();
212 foreach($changes as $rev) {
213 $n = 0;
214 while (!is_null($rev) && $rev['date'] >= $lastupdate &&
215 ($_SERVER['REMOTE_USER'] === $rev['user'] ||
216 $rev['type'] === DOKU_CHANGE_TYPE_MINOR_EDIT)) {
217 $rev = getRevisions($rev['id'], $n++, 1);
218 $rev = (count($rev) > 0) ? $rev[0] : null;
219 }
220
221 if (!is_null($rev) && $rev['date'] >= $lastupdate) {
222 // Some change was not a minor one and not by myself
223 $change_ids[] = $rev['id'];
224 }
225 }
226
227 if ($style === 'digest') {
228 foreach($change_ids as $change_id) {
229 subscription_send_digest($USERINFO['mail'], $change_id,
230 $lastupdate);
231 }
232 } elseif ($style === 'list') {
233 subscription_send_list($USERINFO['mail'], $change_ids, $id);
234 }
235 // TODO: Handle duplicate subscriptions.
236
237 // Update notification time.
238 subscription_set($user, $id, $style, time(), true);
239 }
240 subscription_unlock($id);
241 }
242
243 // restore current user info
244 $USERINFO = $olduinfo;
245 $_SERVER['REMOTE_USER'] = $olduser;
246}
247
248/**
249 * Just send a 1x1 pixel blank gif to the browser
250 *
251 * @author Andreas Gohr <[email protected]>
252 * @author Harry Fuecks <[email protected]>
253 */
254function sendGIF(){
255 if(isset($_REQUEST['debug'])){
256 header('Content-Type: text/plain');
257 return;
258 }
259 $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7');
260 header('Content-Type: image/gif');
261 header('Content-Length: '.strlen($img));
262 header('Connection: Close');
263 print $img;
264 flush();
265 // Browser should drop connection after this
266 // Thinks it's got the whole image
267}
268
269//Setup VIM: ex: et ts=4 :
270// No trailing PHP closing tag - no output please!
271// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php
Note: See TracBrowser for help on using the repository browser.