rebuildtextindex.php

Go to the documentation of this file.

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

class RebuildTextIndex extends Maintenance {
    const RTI_CHUNK_SIZE = 500;

    private $db;

    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
        $this->db = $this->getDB( DB_MASTER );
        if ( $this->db->getType() == 'postgres' ) {
            $this->error( "This script is not needed when using Postgres.\n", true );
        }

        if ( $this->db->getType() == 'sqlite' ) {
            if ( !DatabaseSqlite::getFulltextSearchModule() ) {
                $this->error( "Your version of SQLite module for PHP doesn't "
                    . "support full-text search (FTS3).\n", true );
            }
            if ( !$this->db->checkForEnabledSearch() ) {
                $this->error( "Your database schema is not configured for "
                    . "full-text search support. Run update.php.\n", true );
            }
        }

        if ( $this->db->getType() == 'mysql' ) {
            $this->dropMysqlTextIndex();
            $this->clearSearchIndex();
            $this->populateSearchIndex();
            $this->createMysqlTextIndex();
        } else {
            $this->clearSearchIndex();
            $this->populateSearchIndex();
        }

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

    protected function populateSearchIndex() {
        $res = $this->db->select( 'page', 'MAX(page_id) AS count' );
        $s = $this->db->fetchObject( $res );
        $count = $s->count;
        $this->output( "Rebuilding index fields for {$count} pages...\n" );
        $n = 0;

        $fields = array_merge(
            Revision::selectPageFields(),
            Revision::selectFields(),
            Revision::selectTextFields()
        );

        while ( $n < $count ) {
            if ( $n ) {
                $this->output( $n . "\n" );
            }
            $end = $n + self::RTI_CHUNK_SIZE - 1;

            $res = $this->db->select( [ 'page', 'revision', 'text' ], $fields,
                [ "page_id BETWEEN $n AND $end", 'page_latest = rev_id', 'rev_text_id = old_id' ],
                __METHOD__
            );

            foreach ( $res as $s ) {
                try {
                    $title = Title::makeTitle( $s->page_namespace, $s->page_title );

                    $rev = new Revision( $s );
                    $content = $rev->getContent();

                    $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() {
        $searchindex = $this->db->tableName( 'searchindex' );
        if ( $this->db->indexExists( 'searchindex', 'si_title', __METHOD__ ) ) {
            $this->output( "Dropping index...\n" );
            $sql = "ALTER TABLE $searchindex DROP INDEX si_title, DROP INDEX si_text";
            $this->db->query( $sql, __METHOD__ );
        }
    }

    private function createMysqlTextIndex() {
        $searchindex = $this->db->tableName( 'searchindex' );
        $this->output( "\nRebuild the index...\n" );
        $sql = "ALTER TABLE $searchindex ADD FULLTEXT si_title (si_title), " .
            "ADD FULLTEXT si_text (si_text)";
        $this->db->query( $sql, __METHOD__ );
    }

    private function clearSearchIndex() {
        $this->output( 'Clearing searchindex table...' );
        $this->db->delete( 'searchindex', '*', __METHOD__ );
        $this->output( "Done\n" );
    }
}

$maintClass = "RebuildTextIndex";
require_once RUN_MAINTENANCE_IF_MAIN;