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  * @ingroup Maintenance
  13  */
  14
  15 require_once( "Maintenance.php" );
  16
  17 class UpdateSearchIndex extends Maintenance {
  18
  19         public function __construct() {
  20                 parent::__construct();
  21                 $this->mDescription = "Script for periodic off-peak updating of the search index";
  22                 $this->addParam( 's', 'starting timestamp', false, true );
  23                 $this->addParam( 'e', 'Ending timestamp', false, true );
  24                 $this->addParam( 'p', 'File for saving/loading timestamps, searchUpdate.WIKI_ID.pos by default', false, true );
  25                 $this->addParam( 'l', 'How long the searchindex and revision tables will be locked for', false, true );
  26         }
  27
  28         public function execute() {
  29                 $posFile = $this->getOption( 'p', 'searchUpdate.' . wfWikiId() . '.pos' );
  30                 $end = $this->getOption( 'e', wfTimestampNow() );
  31                 if ( $this->hasOption( 's' ) ) {
  32                         $start = $this->getOption('s');
  33                 } elseif( is_readable( 'searchUpdate.pos' ) ) {
  34                         # B/c to the old position file name which was hardcoded
  35                         # We can safely delete the file when we're done though.
  36                         $start = file_get_contents( 'searchUpdate.pos' );
  37                         unlink( 'searchUpdate.pos' );
  38                 } else {
  39                         $start = @file_get_contents( $posFile );
  40                         if ( !$start ) {
  41                                 $start = wfTimestamp( TS_MW, time() - 86400 );
  42                         }
  43                 }
  44                 $lockTime = $this->getOption( 'l', 20 );
  45
  46                 $this->updateSearchIndex( $start, $end, $lockTime );
  47                 $file = fopen( $posFile, 'w' );
  48                 fwrite( $file, $end );
  49                 fclose( $file );
  50         }
  51
  52         private function updateSearchIndex( $start, $end, $maxLockTime ) {
  53                 global $wgDisableSearchUpdate;
  54
  55                 $wgDisableSearchUpdate = false;
  56
  57                 $dbw = wfGetDB( DB_MASTER );
  58                 $recentchanges = $dbw->tableName( 'recentchanges' );
  59
  60                 $this->output( "Updating searchindex between $start and $end\n" );
  61
  62                 # Select entries from recentchanges which are on top and between the specified times
  63                 $start = $dbw->strencode( $start );
  64                 $end = $dbw->strencode( $end );
  65
  66                 $page = $dbw->tableName( 'page' );
  67                 $sql = "SELECT rc_cur_id,rc_type,rc_moved_to_ns,rc_moved_to_title FROM $recentchanges
  68                   JOIN $page ON rc_cur_id=page_id AND rc_this_oldid=page_latest
  69                   WHERE rc_timestamp BETWEEN '$start' AND '$end'
  70                   ";
  71                 $res = $dbw->query( $sql, __METHOD__ );
  72
  73
  74                 # Lock searchindex
  75                 if ( $maxLockTime ) {
  76                         $this->output( "   --- Waiting for lock ---" );
  77                         $this->lockSearchindex( $dbw );
  78                         $lockTime = time();
  79                         $this->output( "\n" );
  80                 }
  81
  82                 # Loop through the results and do a search update
  83                 while ( $row = $dbw->fetchObject( $res ) ) {
  84                         # Allow reads to be processed
  85                         if ( $maxLockTime && time() > $lockTime + $maxLockTime ) {
  86                                 $this->output( "    --- Relocking ---" );
  87                                 $this->relockSearchindex( $dbw );
  88                                 $lockTime = time();
  89                                 $this->output( "\n" );
  90                         }
  91                         if ( $row->rc_type == RC_LOG ) {
  92                                 continue;
  93                         } elseif ( $row->rc_type == RC_MOVE || $row->rc_type == RC_MOVE_OVER_REDIRECT ) {
  94                                 # Rename searchindex entry
  95                                 $titleObj = Title::makeTitle( $row->rc_moved_to_ns, $row->rc_moved_to_title );
  96                                 $title = $titleObj->getPrefixedDBkey();
  97                                 $this->output( "$title..." );
  98                                 $u = new SearchUpdate( $row->rc_cur_id, $title, false );
  99                                 $this->output( "\n" );
 100                         } else {
 101                                 // Get current revision
 102                                 $rev = Revision::loadFromPageId( $dbw, $row->rc_cur_id );
 103                                 if( $rev ) {
 104                                         $titleObj = $rev->getTitle();
 105                                         $title = $titleObj->getPrefixedDBkey();
 106                                         $this->output( $title );
 107                                         # Update searchindex
 108                                         $u = new SearchUpdate( $row->rc_cur_id, $titleObj->getText(), $rev->getText() );
 109                                         $u->doUpdate();
 110                                         $this->output( "\n" );
 111                                 }
 112                         }
 113                 }
 114
 115                 # Unlock searchindex
 116                 if ( $maxLockTime ) {
 117                         $this->output( "    --- Unlocking --" );
 118                         $this->unlockSearchindex( $dbw );
 119                         $this->output( "\n" );
 120                 }
 121                 $this->output( "Done\n" );
 122         }
 123
 124         /**
 125          * Lock the search index
 126          * @param &$db Database object
 127          */
 128         private function lockSearchindex( &$db ) {
 129                 $write = array( 'searchindex' );
 130                 $read = array( 'page', 'revision', 'text', 'interwiki' );
 131                 $items = array();
 132
 133                 foreach( $write as $table ) {
 134                         $items[] = $db->tableName( $table ) . ' LOW_PRIORITY WRITE';
 135                 }
 136                 foreach( $read as $table ) {
 137                         $items[] = $db->tableName( $table ) . ' READ';
 138                 }
 139                 $sql = "LOCK TABLES " . implode( ',', $items );
 140                 $db->query( $sql, 'updateSearchIndex.php ' . __METHOD__ );
 141         }
 142
 143         /**
 144          * Unlock the tables
 145          * @param &$db Database object
 146          */
 147         private function unlockSearchindex( &$db ) {
 148                 $db->query( "UNLOCK TABLES", 'updateSearchIndex.php ' . __METHOD__ );
 149         }
 150
 151         /**
 152          * Unlock and lock again
 153          * Since the lock is low-priority, queued reads will be able to complete
 154          * @param &$db Database object
 155          */
 156         private function relockSearchindex( &$db ) {
 157                 $this->unlockSearchindex( $db );
 158                 $this->lockSearchindex( $db );
 159         }
 160 }
 161
 162 $maintClass = "UpdateSearchIndex";
 163 require_once( DO_MAINTENANCE );