rebuildtextindex.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\Deferred\SearchUpdate;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Title\Title;
use Wikimedia\Rdbms\DatabaseSqlite;
 
class RebuildTextIndex extends Maintenance {
    private const RTI_CHUNK_SIZE = 500;
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Rebuild search index table from scratch' );
    }
 
    public function getDbType() {
        return Maintenance::DB_ADMIN;
    }
 
    public function execute() {
        // Shouldn't be needed for Postgres
        $dbw = $this->getPrimaryDB();
        if ( $dbw->getType() == 'postgres' ) {
            $this->fatalError( "This script is not needed when using Postgres.\n" );
        }
 
        if ( $dbw->getType() == 'sqlite' ) {
            if ( !DatabaseSqlite::getFulltextSearchModule() ) {
                $this->fatalError( "Your version of SQLite module for PHP doesn't "
                    . "support full-text search (FTS3).\n" );
            }
        }
 
        if ( $dbw->getType() == 'mysql' ) {
            $this->dropMysqlTextIndex();
            $this->clearSearchIndex();
            $this->populateSearchIndex();
            $this->createMysqlTextIndex();
        } else {
            $this->clearSearchIndex();
            $this->populateSearchIndex();
        }
 
        $this->output( "Done.\n" );
    }
 
    protected function populateSearchIndex() {
        $dbw = $this->getPrimaryDB();
        $res = $dbw->newSelectQueryBuilder()
            ->select( [ 'count' => 'MAX(page_id)' ] )
            ->from( 'page' )
            ->caller( __METHOD__ )->fetchResultSet();
        $s = $res->fetchObject();
        $count = $s->count;
        $this->output( "Rebuilding index fields for {$count} pages...\n" );
        $n = 0;
 
        $revStore = $this->getServiceContainer()->getRevisionStore();
        $queryBuilderTemplate = $revStore->newSelectQueryBuilder( $dbw )
            ->joinPage()
            ->joinComment();
 
        while ( $n < $count ) {
            if ( $n ) {
                $this->output( $n . "\n" );
            }
            $end = $n + self::RTI_CHUNK_SIZE - 1;
            $queryBuilder = clone $queryBuilderTemplate;
            $res = $queryBuilder->where( [
                    $dbw->expr( 'page_id', '>=', $n )->and( 'page_id', '<=', $end ),
                    'page_latest = rev_id'
                ] )->caller( __METHOD__ )->fetchResultSet();
 
            foreach ( $res as $s ) {
 
                // T268673 Prevent failure of WikiPage.php: Invalid or virtual namespace -1 given
                if ( $s->page_namespace < 0 ) {
                    continue;
                }
 
                $title = Title::makeTitle( $s->page_namespace, $s->page_title );
                try {
                    $revRecord = $revStore->newRevisionFromRow( $s );
                    $content = $revRecord->getContent( SlotRecord::MAIN );
 
                    $u = new SearchUpdate( $s->page_id, $title, $content );
                    $u->doUpdate();
                } catch ( MWContentSerializationException $ex ) {
                    $this->output( "Failed to deserialize content of revision {$s->rev_id} of page "
                        . "`" . $title->getPrefixedDBkey() . "`!\n" );
                }
            }
            $n += self::RTI_CHUNK_SIZE;
        }
    }
 
    private function dropMysqlTextIndex() {
        $dbw = $this->getDB( DB_PRIMARY );
        $searchindex = $dbw->tableName( 'searchindex' );
        if ( $dbw->indexExists( 'searchindex', 'si_title', __METHOD__ ) ) {
            $this->output( "Dropping index...\n" );
            $sql = "ALTER TABLE $searchindex DROP INDEX si_title, DROP INDEX si_text";
            $dbw->query( $sql, __METHOD__ );
        }
    }
 
    private function createMysqlTextIndex() {
        $dbw = $this->getPrimaryDB();
        $searchindex = $dbw->tableName( 'searchindex' );
        $this->output( "\nRebuild the index...\n" );
        foreach ( [ 'si_title', 'si_text' ] as $field ) {
            $sql = "ALTER TABLE $searchindex ADD FULLTEXT $field ($field)";
            $dbw->query( $sql, __METHOD__ );
        }
    }
 
    private function clearSearchIndex() {
        $dbw = $this->getPrimaryDB();
        $this->output( 'Clearing searchindex table...' );
        $dbw->newDeleteQueryBuilder()
            ->deleteFrom( 'searchindex' )
            ->where( '*' )
            ->caller( __METHOD__ )->execute();
        $this->output( "Done\n" );
    }
}
 
$maintClass = RebuildTextIndex::class;
require_once RUN_MAINTENANCE_IF_MAIN;