fixInconsistentRedirects.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
class FixInconsistentRedirects extends LoggedUpdateMaintenance {
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Fix redirect pages with missing or incomplete row in the redirect table' );
        $this->setBatchSize( 100 );
    }
 
    protected function getUpdateKey() {
        return __CLASS__;
    }
 
    protected function doDBUpdates() {
        $dbr = $this->getReplicaDB();
 
        $builder = $dbr->newSelectQueryBuilder()
            ->from( 'page' )
            ->where( [ 'page_is_redirect' => 1 ] );
 
        $this->output( "Fixing inconsistent redirects ...\n" );
 
        $estimateCount = $builder->estimateRowCount();
        $this->output( "Estimated redirect page count: $estimateCount\n" );
 
        $builder
            ->limit( $this->getBatchSize() )
            ->leftJoin( 'redirect', null, 'page_id=rd_from' )
            ->select( [ 'rd_from', 'rd_interwiki', 'rd_fragment' ] );
 
        // Using the page_redirect_namespace_len index to skip non-redirects
        $index = [ 'page_is_redirect', 'page_namespace', 'page_len', 'page_id' ];
        $builder->select( $index )->orderBy( $index );
        $prevRow = [];
 
        $total = 0;
        $updated = 0;
        do {
            $res = ( clone $builder )
                ->where( $prevRow ? [ $dbr->buildComparison( '>', $prevRow ) ] : [] )
                ->caller( __METHOD__ )->fetchResultSet();
 
            foreach ( $res as $row ) {
                // Only attempt write queries if the row or rd_interwiki/rd_fragment fields are missing
                // (we don't include this condition in the query to avoid slow queries and bad estimates)
                if ( $row->rd_from === null || $row->rd_interwiki === null || $row->rd_fragment === null ) {
                    RefreshLinks::fixRedirect( $this, $row->page_id );
                    $updated++;
                }
            }
            if ( isset( $row ) ) {
                // Update the conditions to select the next batch
                foreach ( $index as $field ) {
                    $prevRow[ $field ] = $row->$field;
                }
            }
 
            $this->waitForReplication();
            $total += $res->numRows();
            $this->output( "$updated/$total\n" );
 
        } while ( $res->numRows() == $this->getBatchSize() );
 
        $this->output( "Done, updated $updated of $total rows.\n" );
        return true;
    }
}
 
$maintClass = FixInconsistentRedirects::class;
require_once RUN_MAINTENANCE_IF_MAIN;