migrateLinksTable.php

Go to the documentation of this file.

<?php
 
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\Title\TitleValue;
 
class MigrateLinksTable extends LoggedUpdateMaintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription(
            'Populates normalization column in links tables.'
        );
        $this->addOption(
            'table',
            'Table name. Like templatelinks.',
            true,
            true
        );
        $this->addOption(
            'sleep',
            'Sleep time (in seconds) between every batch. Default: 0',
            false,
            true
        );
        $this->setBatchSize( 1000 );
    }
 
    protected function getUpdateKey() {
        return __CLASS__ . $this->getOption( 'table', '' );
    }
 
    protected function doDBUpdates() {
        $dbw = $this->getDB( DB_PRIMARY );
        $mapping = \MediaWiki\Linker\LinksMigration::$mapping;
        $table = $this->getOption( 'table', '' );
        if ( !isset( $mapping[$table] ) ) {
            $this->output( "Mapping for this table doesn't exist yet.\n" );
            return false;
        }
        $targetColumn = $mapping[$table]['target_id'];
        if ( !$dbw->fieldExists( $table, $mapping[$table]['title'], __METHOD__ ) ) {
            $this->output( "Old fields don't exist. There is no need to run this script\n" );
            return true;
        }
        if ( !$dbw->fieldExists( $table, $targetColumn, __METHOD__ ) ) {
            $this->output( "Run update.php to create the $targetColumn column.\n" );
            return false;
        }
        if ( !$dbw->tableExists( 'linktarget', __METHOD__ ) ) {
            $this->output( "Run update.php to create the linktarget table.\n" );
            return true;
        }
 
        $this->output( "Populating the $targetColumn column\n" );
        $updated = 0;
 
        $highestPageId = $dbw->newSelectQueryBuilder()
            ->select( 'page_id' )
            ->from( 'page' )
            ->limit( 1 )
            ->caller( __METHOD__ )
            ->orderBy( 'page_id', 'DESC' )
            ->fetchResultSet()->fetchRow();
        if ( !$highestPageId ) {
            $this->output( "Page table is empty.\n" );
            return true;
        }
        $highestPageId = $highestPageId[0];
        $pageId = 0;
        while ( $pageId <= $highestPageId ) {
            // Given the indexes and the structure of links tables,
            // we need to split the update into batches of pages.
            // Otherwise the queries will take a really long time in production and cause read-only.
            $updated += $this->handlePageBatch( $pageId, $mapping, $table );
            $pageId += $this->getBatchSize();
        }
 
        $this->output( "Completed normalization of $table, $updated rows updated.\n" );
 
        return true;
    }
 
    private function handlePageBatch( $lowPageId, $mapping, $table ) {
        $batchSize = $this->getBatchSize();
        $targetColumn = $mapping[$table]['target_id'];
        $pageIdColumn = $mapping[$table]['page_id'];
        // range is inclusive, let's subtract one.
        $highPageId = $lowPageId + $batchSize - 1;
        $dbw = $this->getPrimaryDB();
        $updated = 0;
 
        while ( true ) {
            $res = $dbw->newSelectQueryBuilder()
                ->select( [ $mapping[$table]['ns'], $mapping[$table]['title'] ] )
                ->from( $table )
                ->where( [
                    $targetColumn => [ null, 0 ],
                    $dbw->expr( $pageIdColumn, '>=', $lowPageId ),
                    $dbw->expr( $pageIdColumn, '<=', $highPageId ),
                ] )
                ->limit( 1 )
                ->caller( __METHOD__ )
                ->fetchResultSet();
            if ( !$res->numRows() ) {
                break;
            }
            $row = $res->fetchRow();
            $ns = $row[$mapping[$table]['ns']];
            $titleString = $row[$mapping[$table]['title']];
            $title = new TitleValue( (int)$ns, $titleString );
            $this->output( "Starting backfill of $ns:$titleString " .
                "title on pages between $lowPageId and $highPageId\n" );
            $id = $this->getServiceContainer()->getLinkTargetLookup()->acquireLinkTargetId( $title, $dbw );
            $dbw->newUpdateQueryBuilder()
                ->update( $table )
                ->set( [ $targetColumn => $id ] )
                ->where( [
                    $targetColumn => [ null, 0 ],
                    $mapping[$table]['ns'] => $ns,
                    $mapping[$table]['title'] => $titleString,
                    $dbw->expr( $pageIdColumn, '>=', $lowPageId ),
                    $dbw->expr( $pageIdColumn, '<=', $highPageId ),
                ] )
                ->caller( __METHOD__ )->execute();
            $updatedInThisBatch = $dbw->affectedRows();
            $updated += $updatedInThisBatch;
            $this->output( "Updated $updatedInThisBatch rows\n" );
            // Sleep between batches for replication to catch up
            $this->waitForReplication();
            $sleep = (int)$this->getOption( 'sleep', 0 );
            if ( $sleep > 0 ) {
                sleep( $sleep );
            }
        }
        return $updated;
    }
 
}
 
$maintClass = MigrateLinksTable::class;
require_once RUN_MAINTENANCE_IF_MAIN;