cleanupTitles.php

Go to the documentation of this file.

<?php
use MediaWiki\Title\Title;
 
require_once __DIR__ . '/TableCleanup.php';
 
class TitleCleanup extends TableCleanup {
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Script to clean up broken, unparseable titles' );
        $this->setBatchSize( 1000 );
    }
 
    protected function processRow( $row ) {
        $display = Title::makeName( $row->page_namespace, $row->page_title );
        $verified = $this->getServiceContainer()->getContentLanguage()->normalize( $display );
        $title = Title::newFromText( $verified );
 
        if ( $title !== null
            && $title->canExist()
            && $title->getNamespace() == $row->page_namespace
            && $title->getDBkey() === $row->page_title
        ) {
            // all is fine
            $this->progress( 0 );
 
            return;
        }
 
        if ( $row->page_namespace == NS_FILE && $this->fileExists( $row->page_title ) ) {
            $this->output( "file $row->page_title needs cleanup, please run cleanupImages.php.\n" );
            $this->progress( 0 );
        } elseif ( $title === null ) {
            $this->output( "page $row->page_id ($display) is illegal.\n" );
            $this->moveIllegalPage( $row );
            $this->progress( 1 );
        } else {
            $this->output( "page $row->page_id ($display) doesn't match self.\n" );
            $this->moveInconsistentPage( $row, $title );
            $this->progress( 1 );
        }
    }
 
    protected function fileExists( $name ) {
        // XXX: Doesn't actually check for file existence, just presence of image record.
        // This is reasonable, since cleanupImages.php only iterates over the image table.
        $dbr = $this->getReplicaDB();
        $row = $dbr->newSelectQueryBuilder()
            ->select( '*' )
            ->from( 'image' )
            ->where( [ 'img_name' => $name ] )
            ->caller( __METHOD__ )
            ->fetchRow();
 
        return $row !== false;
    }
 
    protected function moveIllegalPage( $row ) {
        $legal = 'A-Za-z0-9_/\\\\-';
        $legalized = preg_replace_callback( "!([^$legal])!",
            [ $this, 'hexChar' ],
            $row->page_title );
        if ( $legalized == '.' ) {
            $legalized = '(dot)';
        }
        if ( $legalized == '_' ) {
            $legalized = '(space)';
        }
        $legalized = 'Broken/' . $legalized;
 
        $title = Title::newFromText( $legalized );
        if ( $title === null ) {
            $clean = 'Broken/id:' . $row->page_id;
            $this->output( "Couldn't legalize; form '$legalized' still invalid; using '$clean'\n" );
            $title = Title::newFromText( $clean );
        } elseif ( $title->exists() ) {
            $clean = 'Broken/id:' . $row->page_id;
            $this->output( "Legalized for '$legalized' exists; using '$clean'\n" );
            $title = Title::newFromText( $clean );
        }
 
        $dest = $title->getDBkey();
        if ( $this->dryrun ) {
            $this->output( "DRY RUN: would rename $row->page_id ($row->page_namespace," .
                "'$row->page_title') to ($row->page_namespace,'$dest')\n" );
        } else {
            $this->output( "renaming $row->page_id ($row->page_namespace," .
                "'$row->page_title') to ($row->page_namespace,'$dest')\n" );
            $this->getPrimaryDB()
                ->newUpdateQueryBuilder()
                ->update( 'page' )
                ->set( [ 'page_title' => $dest ] )
                ->where( [ 'page_id' => $row->page_id ] )
                ->caller( __METHOD__ )->execute();
        }
    }
 
    protected function moveInconsistentPage( $row, Title $title ) {
        if ( $title->exists( IDBAccessObject::READ_LATEST )
            || $title->getInterwiki()
            || !$title->canExist()
        ) {
            $titleImpossible = $title->getInterwiki() || !$title->canExist();
            if ( $titleImpossible ) {
                $prior = $title->getPrefixedDBkey();
            } else {
                $prior = $title->getDBkey();
            }
 
            # Old cleanupTitles could move articles there. See T25147.
            $ns = $row->page_namespace;
            if ( $ns < 0 ) {
                $ns = 0;
            }
 
            # Namespace which no longer exists. Put the page in the main namespace
            # since we don't have any idea of the old namespace name. See T70501.
            if ( !$this->getServiceContainer()->getNamespaceInfo()->exists( $ns ) ) {
                $ns = 0;
            }
 
            if ( !$titleImpossible && !$title->exists() ) {
                // Looks like the current title, after cleaning it up, is valid and available
                $clean = $prior;
            } else {
                $clean = 'Broken/' . $prior;
            }
            $verified = Title::makeTitleSafe( $ns, $clean );
            if ( !$verified || $verified->exists() ) {
                $blah = "Broken/id:" . $row->page_id;
                $this->output( "Couldn't legalize; form '$clean' exists; using '$blah'\n" );
                $verified = Title::makeTitleSafe( $ns, $blah );
            }
            $title = $verified;
        }
        if ( $title === null ) {
            $this->fatalError( "Something awry; empty title." );
        }
        $ns = $title->getNamespace();
        $dest = $title->getDBkey();
 
        if ( $this->dryrun ) {
            $this->output( "DRY RUN: would rename $row->page_id ($row->page_namespace," .
                "'$row->page_title') to ($ns,'$dest')\n" );
        } else {
            $this->output( "renaming $row->page_id ($row->page_namespace," .
                "'$row->page_title') to ($ns,'$dest')\n" );
            $this->getPrimaryDB()
                ->newUpdateQueryBuilder()
                ->update( 'page' )
                ->set( [
                    'page_namespace' => $ns,
                    'page_title' => $dest
                ] )
                ->where( [ 'page_id' => $row->page_id ] )
                ->caller( __METHOD__ )->execute();
            $this->getServiceContainer()->getLinkCache()->clear();
        }
    }
}
 
$maintClass = TitleCleanup::class;
require_once RUN_MAINTENANCE_IF_MAIN;