updateSearchIndex.php

Go to the documentation of this file.

<?php
require_once( __DIR__ . '/Maintenance.php' );

class UpdateSearchIndex extends Maintenance {

        public function __construct() {
                parent::__construct();
                $this->mDescription = "Script for periodic off-peak updating of the search index";
                $this->addOption( 's', 'starting timestamp', false, true );
                $this->addOption( 'e', 'Ending timestamp', false, true );
                $this->addOption( 'p', 'File for saving/loading timestamps, searchUpdate.WIKI_ID.pos by default', false, true );
                $this->addOption( 'l', 'How long the searchindex and revision tables will be locked for', false, true );
        }

        public function getDbType() {
                return Maintenance::DB_ADMIN;
        }

        public function execute() {
                $posFile = $this->getOption( 'p', 'searchUpdate.' . wfWikiId() . '.pos' );
                $end = $this->getOption( 'e', wfTimestampNow() );
                if ( $this->hasOption( 's' ) ) {
                        $start = $this->getOption( 's' );
                } elseif ( is_readable( 'searchUpdate.pos' ) ) {
                        # B/c to the old position file name which was hardcoded
                        # We can safely delete the file when we're done though.
                        $start = file_get_contents( 'searchUpdate.pos' );
                        unlink( 'searchUpdate.pos' );
                } elseif( is_readable( $posFile ) ) {
                        $start = file_get_contents( $posFile );
                } else {
                        $start = wfTimestamp( TS_MW, time() - 86400 );
                }
                $lockTime = $this->getOption( 'l', 20 );

                $this->doUpdateSearchIndex( $start, $end, $lockTime );
                if ( is_writable( dirname( realpath( $posFile ) ) ) ) {
                        $file = fopen( $posFile, 'w' );
                        if ( $file !== false ) {
                                fwrite( $file, $end );
                                fclose( $file );
                        } else {
                                $this->error( "*** Couldn't write to the $posFile!\n" );
                        }
                } else {
                        $this->error( "*** Couldn't write to the $posFile!\n" );
                }
        }

        private function doUpdateSearchIndex( $start, $end, $maxLockTime ) {
                global $wgDisableSearchUpdate;

                $wgDisableSearchUpdate = false;

                $dbw = wfGetDB( DB_MASTER );
                $recentchanges = $dbw->tableName( 'recentchanges' );

                $this->output( "Updating searchindex between $start and $end\n" );

                # Select entries from recentchanges which are on top and between the specified times
                $start = $dbw->timestamp( $start );
                $end = $dbw->timestamp( $end );

                $page = $dbw->tableName( 'page' );
                $sql = "SELECT rc_cur_id,rc_type,rc_moved_to_ns,rc_moved_to_title FROM $recentchanges
                  JOIN $page ON rc_cur_id=page_id AND rc_this_oldid=page_latest
                  WHERE rc_timestamp BETWEEN '$start' AND '$end'
                  ";
                $res = $dbw->query( $sql, __METHOD__ );

                $this->updateSearchIndex( $maxLockTime, array( $this, 'searchIndexUpdateCallback' ), $dbw, $res );

                $this->output( "Done\n" );
        }

        public function searchIndexUpdateCallback( $dbw, $row ) {
                if ( $row->rc_type == RC_MOVE || $row->rc_type == RC_MOVE_OVER_REDIRECT ) {
                        # Rename searchindex entry
                        $titleObj = Title::makeTitle( $row->rc_moved_to_ns, $row->rc_moved_to_title );
                        $title = $titleObj->getPrefixedDBkey();
                        $this->output( "$title..." );
                        $u = new SearchUpdate( $row->rc_cur_id, $title, false );
                        $u->doUpdate();
                        $this->output( "\n" );
                } elseif ( $row->rc_type !== RC_LOG ) {
                        $this->updateSearchIndexForPage( $dbw, $row->rc_cur_id );
                }
        }
}

$maintClass = "UpdateSearchIndex";
require_once( RUN_MAINTENANCE_IF_MAIN );