updateDoubleWidthSearch.php

Go to the documentation of this file.

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

class UpdateDoubleWidthSearch extends Maintenance {

    public function __construct() {
        parent::__construct();
        $this->mDescription = "Script to normalize double-byte latin UTF-8 characters";
        $this->addOption( 'q', 'quiet', 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() {
        $maxLockTime = $this->getOption( 'l', 20 );

        $dbw = wfGetDB( DB_MASTER );
        if ( $dbw->getType() !== 'mysql' ) {
            $this->error( "This change is only needed on MySQL, quitting.\n", true );
        }

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

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

    public function searchIndexUpdateCallback( $dbw, $row ) {
        return $this->updateSearchIndexForPage( $dbw, $row->si_page );
    }

    private function findRows( $dbw ) {
        $searchindex = $dbw->tableName( 'searchindex' );
        $regexp = '[[:<:]]u8efbd([89][1-9a]|8[b-f]|90)[[:>:]]';
        $sql = "SELECT si_page FROM $searchindex
                 WHERE ( si_text RLIKE '$regexp' )
                    OR ( si_title RLIKE '$regexp' )";

        return $dbw->query( $sql, __METHOD__ );
    }
}

$maintClass = "UpdateDoubleWidthSearch";
require_once RUN_MAINTENANCE_IF_MAIN;