updateSearchIndex.php

Go to the documentation of this file.

<?php
use MediaWiki\Maintenance\Maintenance;
use MediaWiki\RecentChanges\RecentChange;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Search\SearchUpdate;
use MediaWiki\Title\Title;
use MediaWiki\WikiMap\WikiMap;
use Wikimedia\Rdbms\IDBAccessObject;
use Wikimedia\Timestamp\TimestampFormat as TS;
 
// @codeCoverageIgnoreStart
require_once __DIR__ . '/Maintenance.php';
// @codeCoverageIgnoreEnd
 
class UpdateSearchIndex extends Maintenance {
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( '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',
            'Deprecated, has no effect (formerly lock time)',
            false,
            true
        );
    }
 
    public function getDbType() {
        return Maintenance::DB_ADMIN;
    }
 
    public function execute() {
        $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
        $posFile = $this->getOption( 'p', 'searchUpdate.' . rawurlencode( $dbDomain ) . '.pos' );
        $end = $this->getOption( 'e', wfTimestampNow() );
        if ( $this->hasOption( 's' ) ) {
            $start = $this->getOption( 's' );
        } elseif ( is_readable( $posFile ) ) {
            $start = file_get_contents( $posFile );
        } else {
            $start = wfTimestamp( TS::MW, time() - 86400 );
        }
 
        $this->doUpdateSearchIndex( $start, $end );
        $file = fopen( $posFile, 'w' );
        if ( $file !== false ) {
            fwrite( $file, $end );
            fclose( $file );
        } else {
            $this->error( "*** Couldn't write to the $posFile!\n" );
        }
    }
 
    private function doUpdateSearchIndex( string $start, string $end ) {
        global $wgDisableSearchUpdate;
 
        $wgDisableSearchUpdate = false;
 
        $dbw = $this->getPrimaryDB();
 
        $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 );
 
        $res = $dbw->newSelectQueryBuilder()
            ->select( 'rc_cur_id' )
            ->from( 'recentchanges' )
            ->join( 'page', null, 'rc_cur_id=page_id AND rc_this_oldid=page_latest' )
            ->where( [
                $dbw->expr( 'rc_source', '!=', RecentChange::SRC_LOG ),
                $dbw->expr( 'rc_timestamp', '>=', $start ),
                $dbw->expr( 'rc_timestamp', '<=', $end ),
            ] )
            ->caller( __METHOD__ )->fetchResultSet();
 
        foreach ( $res as $row ) {
            $this->updateSearchIndexForPage( (int)$row->rc_cur_id );
        }
        $this->output( "Done\n" );
    }
 
    private function updateSearchIndexForPage( int $pageId ) {
        // Get current revision
        $rev = $this->getServiceContainer()
            ->getRevisionLookup()
            ->getRevisionByPageId( $pageId, 0, IDBAccessObject::READ_LATEST );
        $title = null;
        if ( $rev ) {
            $titleObj = Title::newFromPageIdentity( $rev->getPage() );
            $title = $titleObj->getPrefixedDBkey();
            $this->output( "$title..." );
            # Update searchindex
            $u = new SearchUpdate( $pageId, $titleObj, $rev->getContent( SlotRecord::MAIN ) );
            $u->doUpdate();
            $this->output( "\n" );
        }
 
        return $title;
    }
}
 
// @codeCoverageIgnoreStart
$maintClass = UpdateSearchIndex::class;
require_once RUN_MAINTENANCE_IF_MAIN;
// @codeCoverageIgnoreEnd