cleanupImages.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
 
require_once __DIR__ . '/TableCleanup.php';
 
class CleanupImages extends TableCleanup {
    protected $defaultParams = [
        'table' => 'image',
        'conds' => [],
        'index' => 'img_name',
        'callback' => 'processRow',
    ];
 
    private $repo;
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Script to clean up broken, unparseable upload filenames' );
    }
 
    protected function processRow( $row ) {
        $source = $row->img_name;
        if ( $source == '' ) {
            // Ye olde empty rows. Just kill them.
            $this->killRow( $source );
 
            return $this->progress( 1 );
        }
 
        $cleaned = $source;
 
        // About half of old bad image names have percent-codes
        $cleaned = rawurldecode( $cleaned );
 
        // We also have some HTML entities there
        $cleaned = Sanitizer::decodeCharReferences( $cleaned );
 
        $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 
        // Some are old latin-1
        $cleaned = $contLang->checkTitleEncoding( $cleaned );
 
        // Many of remainder look like non-normalized unicode
        $cleaned = $contLang->normalize( $cleaned );
 
        $title = Title::makeTitleSafe( NS_FILE, $cleaned );
 
        if ( $title === null ) {
            $this->output( "page $source ($cleaned) is illegal.\n" );
            $safe = $this->buildSafeTitle( $cleaned );
            if ( $safe === false ) {
                return $this->progress( 0 );
            }
            $this->pokeFile( $source, $safe );
 
            return $this->progress( 1 );
        }
 
        if ( $title->getDBkey() !== $source ) {
            $munged = $title->getDBkey();
            $this->output( "page $source ($munged) doesn't match self.\n" );
            $this->pokeFile( $source, $munged );
 
            return $this->progress( 1 );
        }
 
        return $this->progress( 0 );
    }
 
    private function killRow( $name ) {
        if ( $this->dryrun ) {
            $this->output( "DRY RUN: would delete bogus row '$name'\n" );
        } else {
            $this->output( "deleting bogus row '$name'\n" );
            $db = $this->getDB( DB_PRIMARY );
            $db->delete( 'image',
                [ 'img_name' => $name ],
                __METHOD__ );
        }
    }
 
    private function filePath( $name ) {
        if ( $this->repo === null ) {
            $this->repo = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo();
        }
 
        return $this->repo->getRootDirectory() . '/' . $this->repo->getHashPath( $name ) . $name;
    }
 
    private function imageExists( $name, $db ) {
        return (bool)$db->newSelectQueryBuilder()
            ->select( '1' )
            ->from( 'image' )
            ->where( [ 'img_name' => $name ] )
            ->caller( __METHOD__ )
            ->fetchField();
    }
 
    private function pageExists( $name, $db ) {
        return (bool)$db->newSelectQueryBuilder()
            ->select( '1' )
            ->from( 'page' )
            ->where( [
                'page_namespace' => NS_FILE,
                'page_title' => $name,
            ] )
            ->caller( __METHOD__ )
            ->fetchField();
    }
 
    private function pokeFile( $orig, $new ) {
        $path = $this->filePath( $orig );
        if ( !file_exists( $path ) ) {
            $this->output( "missing file: $path\n" );
            $this->killRow( $orig );
 
            return;
        }
 
        $db = $this->getDB( DB_PRIMARY );
 
        /*
         * To prevent key collisions in the update() statements below,
         * if the target title exists in the image table, or if both the
         * original and target titles exist in the page table, append
         * increasing version numbers until the target title exists in
         * neither.  (See also T18916.)
         */
        $version = 0;
        $final = $new;
        $conflict = ( $this->imageExists( $final, $db ) ||
            ( $this->pageExists( $orig, $db ) && $this->pageExists( $final, $db ) ) );
 
        while ( $conflict ) {
            $this->output( "Rename conflicts with '$final'...\n" );
            $version++;
            $final = $this->appendTitle( $new, "_$version" );
            $conflict = ( $this->imageExists( $final, $db ) || $this->pageExists( $final, $db ) );
        }
 
        $finalPath = $this->filePath( $final );
 
        if ( $this->dryrun ) {
            $this->output( "DRY RUN: would rename $path to $finalPath\n" );
        } else {
            $this->output( "renaming $path to $finalPath\n" );
            // @todo FIXME: Should this use File::move()?
            $this->beginTransaction( $db, __METHOD__ );
            $db->update( 'image',
                [ 'img_name' => $final ],
                [ 'img_name' => $orig ],
                __METHOD__ );
            $db->update( 'oldimage',
                [ 'oi_name' => $final ],
                [ 'oi_name' => $orig ],
                __METHOD__ );
            $db->update( 'page',
                [ 'page_title' => $final ],
                [ 'page_title' => $orig, 'page_namespace' => NS_FILE ],
                __METHOD__ );
            $dir = dirname( $finalPath );
            if ( !file_exists( $dir ) ) {
                if ( !wfMkdirParents( $dir, null, __METHOD__ ) ) {
                    $this->output( "RENAME FAILED, COULD NOT CREATE $dir" );
                    $this->rollbackTransaction( $db, __METHOD__ );
 
                    return;
                }
            }
            if ( rename( $path, $finalPath ) ) {
                $this->commitTransaction( $db, __METHOD__ );
            } else {
                $this->error( "RENAME FAILED" );
                $this->rollbackTransaction( $db, __METHOD__ );
            }
        }
    }
 
    private function appendTitle( $name, $suffix ) {
        return preg_replace( '/^(.*)(\..*?)$/',
            "\\1$suffix\\2", $name );
    }
 
    private function buildSafeTitle( $name ) {
        $x = preg_replace_callback(
            '/([^' . Title::legalChars() . ']|~)/',
            [ $this, 'hexChar' ],
            $name );
 
        $test = Title::makeTitleSafe( NS_FILE, $x );
        if ( $test === null || $test->getDBkey() !== $x ) {
            $this->error( "Unable to generate safe title from '$name', got '$x'" );
 
            return false;
        }
 
        return $x;
    }
}
 
$maintClass = CleanupImages::class;
require_once RUN_MAINTENANCE_IF_MAIN;