maintenance/updateSearchIndex.php

   1 <?php
   2 /**
   3  * Script for periodic off-peak updating of the search index
   4  *
   5  * Usage: php updateSearchIndex.php [-s START] [-e END] [-p POSFILE] [-l LOCKTIME] [-q]
   6  * Where START is the starting timestamp
   7  * END is the ending timestamp
   8  * POSFILE is a file to load timestamps from and save them to, searchUpdate.WIKI_ID.pos by default
   9  * LOCKTIME is how long the searchindex and revision tables will be locked for
  10  * -q means quiet
  11  *
  12  * This program is free software; you can redistribute it and/or modify
  13  * it under the terms of the GNU General Public License as published by
  14  * the Free Software Foundation; either version 2 of the License, or
  15  * (at your option) any later version.
  16  *
  17  * This program is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  20  * GNU General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU General Public License along
  23  * with this program; if not, write to the Free Software Foundation, Inc.,
  24  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  25  * http://www.gnu.org/copyleft/gpl.html
  26  *
  27  * @ingroup Maintenance
  28  */
  29
  30 require_once( dirname(__FILE__) . '/Maintenance.php' );
  31
  32 class UpdateSearchIndex extends Maintenance {
  33
  34         public function __construct() {
  35                 parent::__construct();
  36                 $this->mDescription = "Script for periodic off-peak updating of the search index";
  37                 $this->addOption( 's', 'starting timestamp', false, true );
  38                 $this->addOption( 'e', 'Ending timestamp', false, true );
  39                 $this->addOption( 'p', 'File for saving/loading timestamps, searchUpdate.WIKI_ID.pos by default', false, true );
  40                 $this->addOption( 'l', 'How long the searchindex and revision tables will be locked for', false, true );
  41         }
  42
  43         public function getDbType() {
  44                 return Maintenance::DB_ADMIN;
  45         }
  46
  47         public function execute() {
  48                 $posFile = $this->getOption( 'p', 'searchUpdate.' . wfWikiId() . '.pos' );
  49                 $end = $this->getOption( 'e', wfTimestampNow() );
  50                 if ( $this->hasOption( 's' ) ) {
  51                         $start = $this->getOption('s');
  52                 } elseif( is_readable( 'searchUpdate.pos' ) ) {
  53                         # B/c to the old position file name which was hardcoded
  54                         # We can safely delete the file when we're done though.
  55                         $start = file_get_contents( 'searchUpdate.pos' );
  56                         unlink( 'searchUpdate.pos' );
  57                 } else {
  58                         $start = @file_get_contents( $posFile );
  59                         if ( !$start ) {
  60                                 $start = wfTimestamp( TS_MW, time() - 86400 );
  61                         }
  62                 }
  63                 $lockTime = $this->getOption( 'l', 20 );
  64
  65                 $this->doUpdateSearchIndex( $start, $end, $lockTime );
  66                 $file = fopen( $posFile, 'w' );
  67                 fwrite( $file, $end );
  68                 fclose( $file );
  69         }
  70
  71         private function doUpdateSearchIndex( $start, $end, $maxLockTime ) {
  72                 global $wgDisableSearchUpdate;
  73
  74                 $wgDisableSearchUpdate = false;
  75
  76                 $dbw = wfGetDB( DB_MASTER );
  77                 $recentchanges = $dbw->tableName( 'recentchanges' );
  78
  79                 $this->output( "Updating searchindex between $start and $end\n" );
  80
  81                 # Select entries from recentchanges which are on top and between the specified times
  82                 $start = $dbw->timestamp( $start );
  83                 $end = $dbw->timestamp( $end );
  84
  85                 $page = $dbw->tableName( 'page' );
  86                 $sql = "SELECT rc_cur_id,rc_type,rc_moved_to_ns,rc_moved_to_title FROM $recentchanges
  87                   JOIN $page ON rc_cur_id=page_id AND rc_this_oldid=page_latest
  88                   WHERE rc_timestamp BETWEEN '$start' AND '$end'
  89                   ";
  90                 $res = $dbw->query( $sql, __METHOD__ );
  91
  92
  93                 # Lock searchindex
  94                 if ( $maxLockTime ) {
  95                         $this->output( "   --- Waiting for lock ---" );
  96                         $this->lockSearchindex( $dbw );
  97                         $lockTime = time();
  98                         $this->output( "\n" );
  99                 }
 100
 101                 # Loop through the results and do a search update
 102                 foreach ( $res as $row ) {
 103                         # Allow reads to be processed
 104                         if ( $maxLockTime && time() > $lockTime + $maxLockTime ) {
 105                                 $this->output( "    --- Relocking ---" );
 106                                 $this->relockSearchindex( $dbw );
 107                                 $lockTime = time();
 108                                 $this->output( "\n" );
 109                         }
 110                         if ( $row->rc_type == RC_LOG ) {
 111                                 continue;
 112                         } elseif ( $row->rc_type == RC_MOVE || $row->rc_type == RC_MOVE_OVER_REDIRECT ) {
 113                                 # Rename searchindex entry
 114                                 $titleObj = Title::makeTitle( $row->rc_moved_to_ns, $row->rc_moved_to_title );
 115                                 $title = $titleObj->getPrefixedDBkey();
 116                                 $this->output( "$title..." );
 117                                 $u = new SearchUpdate( $row->rc_cur_id, $title, false );
 118                                 $this->output( "\n" );
 119                         } else {
 120                                 // Get current revision
 121                                 $rev = Revision::loadFromPageId( $dbw, $row->rc_cur_id );
 122                                 if( $rev ) {
 123                                         $titleObj = $rev->getTitle();
 124                                         $title = $titleObj->getPrefixedDBkey();
 125                                         $this->output( $title );
 126                                         # Update searchindex
 127                                         $u = new SearchUpdate( $row->rc_cur_id, $titleObj->getText(), $rev->getText() );
 128                                         $u->doUpdate();
 129                                         $this->output( "\n" );
 130                                 }
 131                         }
 132                 }
 133
 134                 # Unlock searchindex
 135                 if ( $maxLockTime ) {
 136                         $this->output( "    --- Unlocking --" );
 137                         $this->unlockSearchindex( $dbw );
 138                         $this->output( "\n" );
 139                 }
 140                 $this->output( "Done\n" );
 141         }
 142
 143         /**
 144          * Lock the search index
 145          * @param &$db Database object
 146          */
 147         private function lockSearchindex( &$db ) {
 148                 $write = array( 'searchindex' );
 149                 $read = array( 'page', 'revision', 'text', 'interwiki' );
 150                 $db->lockTables( $read, $write, 'updateSearchIndex.php ' . __METHOD__ );
 151         }
 152
 153         /**
 154          * Unlock the tables
 155          * @param &$db Database object
 156          */
 157         private function unlockSearchindex( &$db ) {
 158                 $db->unlockTables( 'updateSearchIndex.php ' . __METHOD__ );
 159         }
 160
 161         /**
 162          * Unlock and lock again
 163          * Since the lock is low-priority, queued reads will be able to complete
 164          * @param &$db Database object
 165          */
 166         private function relockSearchindex( &$db ) {
 167                 $this->unlockSearchindex( $db );
 168                 $this->lockSearchindex( $db );
 169         }
 170 }
 171
 172 $maintClass = "UpdateSearchIndex";
 173 require_once( DO_MAINTENANCE );