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$ID = cleanID($_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 metaUpdate() or 35 runSitemapper() or 36 sendDigest() or 37 runTrimRecentChanges() or 38 runTrimRecentChanges(true) or 39 $evt->advise_after(); 40} 41if($defer) sendGIF(); 42 43if(!$output) ob_end_clean(); 44exit; 45 46// -------------------------------------------------------------------- 47 48/** 49 * Trims the recent changes cache (or imports the old changelog) as needed. 50 * 51 * @param media_changes If the media changelog shall be trimmed instead of 52 * the page changelog 53 * 54 * @author Ben Coburn <btcoburn@silicodon.net> 55 */ 56function runTrimRecentChanges($media_changes = false) { 57 global $conf; 58 59 $fn = ($media_changes ? $conf['media_changelog'] : $conf['changelog']); 60 61 // Trim the Recent Changes 62 // Trims the recent changes cache to the last $conf['changes_days'] recent 63 // changes or $conf['recent'] items, which ever is larger. 64 // The trimming is only done once a day. 65 if (@file_exists($fn) && 66 (@filemtime($fn.'.trimmed')+86400)<time() && 67 !@file_exists($fn.'_tmp')) { 68 @touch($fn.'.trimmed'); 69 io_lock($fn); 70 $lines = file($fn); 71 if (count($lines)<=$conf['recent']) { 72 // nothing to trim 73 io_unlock($fn); 74 return false; 75 } 76 77 io_saveFile($fn.'_tmp', ''); // presave tmp as 2nd lock 78 $trim_time = time() - $conf['recent_days']*86400; 79 $out_lines = array(); 80 81 for ($i=0; $i<count($lines); $i++) { 82 $log = parseChangelogLine($lines[$i]); 83 if ($log === false) continue; // discard junk 84 if ($log['date'] < $trim_time) { 85 $old_lines[$log['date'].".$i"] = $lines[$i]; // keep old lines for now (append .$i to prevent key collisions) 86 } else { 87 $out_lines[$log['date'].".$i"] = $lines[$i]; // definitely keep these lines 88 } 89 } 90 91 if (count($lines)==count($out_lines)) { 92 // nothing to trim 93 @unlink($fn.'_tmp'); 94 io_unlock($fn); 95 return false; 96 } 97 98 // sort the final result, it shouldn't be necessary, 99 // however the extra robustness in making the changelog cache self-correcting is worth it 100 ksort($out_lines); 101 $extra = $conf['recent'] - count($out_lines); // do we need extra lines do bring us up to minimum 102 if ($extra > 0) { 103 ksort($old_lines); 104 $out_lines = array_merge(array_slice($old_lines,-$extra),$out_lines); 105 } 106 107 // save trimmed changelog 108 io_saveFile($fn.'_tmp', implode('', $out_lines)); 109 @unlink($fn); 110 if (!rename($fn.'_tmp', $fn)) { 111 // rename failed so try another way... 112 io_unlock($fn); 113 io_saveFile($fn, implode('', $out_lines)); 114 @unlink($fn.'_tmp'); 115 } else { 116 io_unlock($fn); 117 } 118 return true; 119 } 120 121 // nothing done 122 return false; 123} 124 125/** 126 * Runs the indexer for the current page 127 * 128 * @author Andreas Gohr <andi@splitbrain.org> 129 */ 130function runIndexer(){ 131 global $ID; 132 global $conf; 133 print "runIndexer(): started".NL; 134 135 if(!$ID) return false; 136 137 // check if indexing needed 138 $idxtag = metaFN($ID,'.indexed'); 139 if(@file_exists($idxtag)){ 140 if(trim(io_readFile($idxtag)) == idx_get_version()){ 141 $last = @filemtime($idxtag); 142 if($last > @filemtime(wikiFN($ID))){ 143 print "runIndexer(): index for $ID up to date".NL; 144 return false; 145 } 146 } 147 } 148 149 // try to aquire a lock 150 $lock = $conf['lockdir'].'/_indexer.lock'; 151 while(!@mkdir($lock,$conf['dmode'])){ 152 usleep(50); 153 if(time()-@filemtime($lock) > 60*5){ 154 // looks like a stale lock - remove it 155 @rmdir($lock); 156 print "runIndexer(): stale lock removed".NL; 157 }else{ 158 print "runIndexer(): indexer locked".NL; 159 return false; 160 } 161 } 162 if($conf['dperm']) chmod($lock, $conf['dperm']); 163 164 // do the work 165 idx_addPage($ID); 166 167 // we're finished - save and free lock 168 io_saveFile(metaFN($ID,'.indexed'), idx_get_version()); 169 @rmdir($lock); 170 print "runIndexer(): finished".NL; 171 return true; 172} 173 174/** 175 * Will render the metadata for the page if not exists yet 176 * 177 * This makes sure pages which are created from outside DokuWiki will 178 * gain their data when viewed for the first time. 179 */ 180function metaUpdate(){ 181 global $ID; 182 print "metaUpdate(): started".NL; 183 184 if(!$ID) return false; 185 $file = metaFN($ID, '.meta'); 186 echo "meta file: $file".NL; 187 188 // rendering needed? 189 if (@file_exists($file)) return false; 190 if (!page_exists($ID)) return false; 191 192 global $conf; 193 194 // gather some additional info from changelog 195 $info = io_grep($conf['changelog'], 196 '/^(\d+)\t(\d+\.\d+\.\d+\.\d+)\t'.preg_quote($ID,'/').'\t([^\t]+)\t([^\t\n]+)/', 197 0,true); 198 199 $meta = array(); 200 if(!empty($info)){ 201 $meta['date']['created'] = $info[0][1]; 202 foreach($info as $item){ 203 if($item[4] != '*'){ 204 $meta['date']['modified'] = $item[1]; 205 if($item[3]){ 206 $meta['contributor'][$item[3]] = $item[3]; 207 } 208 } 209 } 210 } 211 212 $meta = p_render_metadata($ID, $meta); 213 p_save_metadata($ID, $meta); 214 215 echo "metaUpdate(): finished".NL; 216 return true; 217} 218 219/** 220 * Builds a Google Sitemap of all public pages known to the indexer 221 * 222 * The map is placed in the root directory named sitemap.xml.gz - This 223 * file needs to be writable! 224 * 225 * @author Andreas Gohr 226 * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html 227 */ 228function runSitemapper(){ 229 print "runSitemapper(): started".NL; 230 $result = Sitemapper::generate() && Sitemapper::pingSearchEngines(); 231 print 'runSitemapper(): finished'.NL; 232 return $result; 233} 234 235/** 236 * Send digest and list mails for all subscriptions which are in effect for the 237 * current page 238 * 239 * @author Adrian Lang <lang@cosmocode.de> 240 */ 241function sendDigest() { 242 echo 'sendDigest(): start'.NL; 243 global $ID; 244 global $conf; 245 if (!$conf['subscribers']) { 246 return; 247 } 248 $subscriptions = subscription_find($ID, array('style' => '(digest|list)', 249 'escaped' => true)); 250 global $auth; 251 global $lang; 252 global $conf; 253 global $USERINFO; 254 255 // remember current user info 256 $olduinfo = $USERINFO; 257 $olduser = $_SERVER['REMOTE_USER']; 258 259 foreach($subscriptions as $id => $users) { 260 if (!subscription_lock($id)) { 261 continue; 262 } 263 foreach($users as $data) { 264 list($user, $style, $lastupdate) = $data; 265 $lastupdate = (int) $lastupdate; 266 if ($lastupdate + $conf['subscribe_time'] > time()) { 267 // Less than the configured time period passed since last 268 // update. 269 continue; 270 } 271 272 // Work as the user to make sure ACLs apply correctly 273 $USERINFO = $auth->getUserData($user); 274 $_SERVER['REMOTE_USER'] = $user; 275 if ($USERINFO === false) { 276 continue; 277 } 278 279 if (substr($id, -1, 1) === ':') { 280 // The subscription target is a namespace 281 $changes = getRecentsSince($lastupdate, null, getNS($id)); 282 } else { 283 if(auth_quickaclcheck($id) < AUTH_READ) continue; 284 285 $meta = p_get_metadata($id); 286 $changes = array($meta['last_change']); 287 } 288 289 // Filter out pages only changed in small and own edits 290 $change_ids = array(); 291 foreach($changes as $rev) { 292 $n = 0; 293 while (!is_null($rev) && $rev['date'] >= $lastupdate && 294 ($_SERVER['REMOTE_USER'] === $rev['user'] || 295 $rev['type'] === DOKU_CHANGE_TYPE_MINOR_EDIT)) { 296 $rev = getRevisions($rev['id'], $n++, 1); 297 $rev = (count($rev) > 0) ? $rev[0] : null; 298 } 299 300 if (!is_null($rev) && $rev['date'] >= $lastupdate) { 301 // Some change was not a minor one and not by myself 302 $change_ids[] = $rev['id']; 303 } 304 } 305 306 if ($style === 'digest') { 307 foreach($change_ids as $change_id) { 308 subscription_send_digest($USERINFO['mail'], $change_id, 309 $lastupdate); 310 } 311 } elseif ($style === 'list') { 312 subscription_send_list($USERINFO['mail'], $change_ids, $id); 313 } 314 // TODO: Handle duplicate subscriptions. 315 316 // Update notification time. 317 subscription_set($user, $id, $style, time(), true); 318 } 319 subscription_unlock($id); 320 } 321 322 // restore current user info 323 $USERINFO = $olduinfo; 324 $_SERVER['REMOTE_USER'] = $olduser; 325} 326 327/** 328 * Just send a 1x1 pixel blank gif to the browser 329 * 330 * @author Andreas Gohr <andi@splitbrain.org> 331 * @author Harry Fuecks <fuecks@gmail.com> 332 */ 333function sendGIF(){ 334 if(isset($_REQUEST['debug'])){ 335 header('Content-Type: text/plain'); 336 return; 337 } 338 $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7'); 339 header('Content-Type: image/gif'); 340 header('Content-Length: '.strlen($img)); 341 header('Connection: Close'); 342 print $img; 343 flush(); 344 // Browser should drop connection after this 345 // Thinks it's got the whole image 346} 347 348//Setup VIM: ex: et ts=4 enc=utf-8 : 349// No trailing PHP closing tag - no output please! 350// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php 351