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',realpath(dirname(__FILE__).'/../../').'/'); 9require_once(DOKU_INC.'inc/init.php'); 10require_once(DOKU_INC.'inc/auth.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// send gif 18sendGIF(); 19 20// Catch any possible output (e.g. errors) 21if(!$_REQUEST['debug']) ob_start(); 22 23// run one of the jobs 24runIndexer() or runSitemapper(); 25 26if(!$_REQUEST['debug']) ob_end_clean(); 27exit; 28 29// -------------------------------------------------------------------- 30 31/** 32 * Runs the indexer for the current page 33 * 34 * @author Andreas Gohr <andi@splitbrain.org> 35 */ 36function runIndexer(){ 37 global $conf; 38 print "runIndexer(): started".NL; 39 40 $ID = cleanID($_REQUEST['id']); 41 if(!$ID) return false; 42 43 // check if indexing needed 44 $last = @filemtime(metaFN($ID,'.indexed')); 45 if($last > @filemtime(wikiFN($ID))){ 46 print "runIndexer(): index for $ID up to date".NL; 47 return false; 48 } 49 50 // try to aquire a lock 51 $lock = $conf['lockdir'].'/_indexer.lock'; 52 while(!@mkdir($lock,0777)){ 53 usleep(50); 54 if(time()-@filemtime($lock) > 60*5){ 55 // looks like a stale lock - remove it 56 @rmdir($lock); 57 print "runIndexer(): stale lock removed".NL; 58 }else{ 59 print "runIndexer(): indexer locked".NL; 60 return false; 61 } 62 } 63 64 require_once(DOKU_INC.'inc/indexer.php'); 65 66 // do the work 67 idx_addPage($ID); 68 69 // we're finished - save and free lock 70 io_saveFile(metaFN($ID,'.indexed'),' '); 71 @rmdir($lock); 72 print "runIndexer(): finished".NL; 73 return true; 74} 75 76/** 77 * Builds a Google Sitemap of all public pages known to the indexer 78 * 79 * The map is placed in the root directory named sitemap.xml.gz - This 80 * file needs to be writable! 81 * 82 * @author Andreas Gohr 83 * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html 84 */ 85function runSitemapper(){ 86 global $conf; 87 print "runSitemapper(): started".NL; 88 if(!$conf['sitemap']) return false; 89 90 if($conf['usegzip']){ 91 $sitemap = DOKU_INC.'sitemap.xml.gz'; 92 }else{ 93 $sitemap = DOKU_INC.'sitemap.xml'; 94 } 95 print "runSitemapper(): using $sitemap".NL; 96 97 if(!is_writable($sitemap)) return false; 98 if(@filesize($sitemap) && 99 @filemtime($sitemap) > (time()-($conf['sitemap']*60*60*24))){ 100 print 'runSitemapper(): Sitemap up to date'.NL; 101 return false; 102 } 103 104 $pages = file($conf['cachedir'].'/page.idx'); 105 print 'runSitemapper(): creating sitemap using '.count($pages).' pages'.NL; 106 107 // build the sitemap 108 ob_start(); 109 print '<?xml version="1.0" encoding="UTF-8"?>'.NL; 110 print '<urlset xmlns="http://www.google.com/schemas/sitemap/0.84">'.NL; 111 foreach($pages as $id){ 112 $id = trim($id); 113 $file = wikiFN($id); 114 115 //skip hidden, non existing and restricted files 116 if(isHiddenPage($id)) return false; 117 $date = @filemtime($file); 118 if(!$date) continue; 119 if(auth_aclcheck($id,'','') < AUTH_READ) continue; 120 121 print ' <url>'.NL; 122 print ' <loc>'.wl($id,'',true).'</loc>'.NL; 123 print ' <lastmod>'.date_iso8601($date).'</lastmod>'.NL; 124 print ' </url>'.NL; 125 } 126 print '</urlset>'.NL; 127 $data = ob_get_contents(); 128 ob_end_clean(); 129 130 //save the new sitemap 131 io_saveFile($sitemap,$data); 132 133 print 'runSitemapper(): pinging google'.NL; 134 //ping google 135 $url = 'http://www.google.com/webmasters/sitemaps/ping?sitemap='; 136 $url .= urlencode(DOKU_URL.$sitemap); 137 $http = new DokuHTTPClient(); 138 $http->get($url); 139 if($http->error) print 'runSitemapper(): '.$http->error.NL; 140 141 print 'runSitemapper(): finished'.NL; 142 return true; 143} 144 145/** 146 * Formats a timestamp as ISO 8601 date 147 * 148 * @author <ungu at terong dot com> 149 * @link http://www.php.net/manual/en/function.date.php#54072 150 */ 151function date_iso8601($int_date) { 152 //$int_date: current date in UNIX timestamp 153 $date_mod = date('Y-m-d\TH:i:s', $int_date); 154 $pre_timezone = date('O', $int_date); 155 $time_zone = substr($pre_timezone, 0, 3).":".substr($pre_timezone, 3, 2); 156 $date_mod .= $time_zone; 157 return $date_mod; 158} 159 160/** 161 * Just send a 1x1 pixel blank gif to the browser 162 * 163 * @author Andreas Gohr <andi@splitbrain.org> 164 * @author Harry Fuecks <fuecks@gmail.com> 165 */ 166function sendGIF(){ 167 if($_REQUEST['debug']){ 168 header('Content-Type: text/plain'); 169 return; 170 } 171 $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7'); 172 header('Content-Type: image/gif'); 173 header('Content-Length: '.strlen($img)); 174 header('Connection: Close'); 175 print $img; 176 flush(); 177 // Browser should drop connection after this 178 // Thinks it's got the whole image 179} 180 181//Setup VIM: ex: et ts=4 enc=utf-8 : 182// No trailing PHP closing tag - no output please! 183// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php 184