findMissingFiles.php

Go to the documentation of this file.

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

class FindMissingFiles extends Maintenance {
    function __construct() {
        parent::__construct();

        $this->addDescription( 'Find registered files with no corresponding file.' );
        $this->addOption( 'start', 'Start after this file name', false, true );
        $this->addOption( 'mtimeafter', 'Only include files changed since this time', false, true );
        $this->addOption( 'mtimebefore', 'Only includes files changed before this time', false, true );
        $this->setBatchSize( 300 );
    }

    function execute() {
        $lastName = $this->getOption( 'start', '' );

        $repo = RepoGroup::singleton()->getLocalRepo();
        $dbr = $repo->getSlaveDB();
        $be = $repo->getBackend();

        $mtime1 = $dbr->timestampOrNull( $this->getOption( 'mtimeafter', null ) );
        $mtime2 = $dbr->timestampOrNull( $this->getOption( 'mtimebefore', null ) );

        $joinTables = [];
        $joinConds = [];
        if ( $mtime1 || $mtime2 ) {
            $joinTables[] = 'page';
            $joinConds['page'] = [ 'INNER JOIN',
                [ 'page_title = img_name', 'page_namespace' => NS_FILE ] ];
            $joinTables[] = 'logging';
            $on = [ 'log_page = page_id', 'log_type' => [ 'upload', 'move', 'delete' ] ];
            if ( $mtime1 ) {
                $on[] = "log_timestamp > {$dbr->addQuotes($mtime1)}";
            }
            if ( $mtime2 ) {
                $on[] = "log_timestamp < {$dbr->addQuotes($mtime2)}";
            }
            $joinConds['logging'] = [ 'INNER JOIN', $on ];
        }

        do {
            $res = $dbr->select(
                array_merge( [ 'image' ], $joinTables ),
                [ 'name' => 'img_name' ],
                [ "img_name > " . $dbr->addQuotes( $lastName ) ],
                __METHOD__,
                // DISTINCT causes a pointless filesort
                [ 'ORDER BY' => 'name', 'GROUP BY' => 'name',
                    'LIMIT' => $this->mBatchSize ],
                $joinConds
            );

            // Check if any of these files are missing...
            $pathsByName = [];
            foreach ( $res as $row ) {
                $file = $repo->newFile( $row->name );
                $pathsByName[$row->name] = $file->getPath();
                $lastName = $row->name;
            }
            $be->preloadFileStat( [ 'srcs' => $pathsByName ] );
            foreach ( $pathsByName as $path ) {
                if ( $be->fileExists( [ 'src' => $path ] ) === false ) {
                    $this->output( "$path\n" );
                }
            }

            // Find all missing old versions of any of the files in this batch...
            if ( count( $pathsByName ) ) {
                $ores = $dbr->select( 'oldimage',
                    [ 'oi_name', 'oi_archive_name' ],
                    [ 'oi_name' => array_keys( $pathsByName ) ],
                    __METHOD__
                );

                $checkPaths = [];
                foreach ( $ores as $row ) {
                    if ( !strlen( $row->oi_archive_name ) ) {
                        continue; // broken row
                    }
                    $file = $repo->newFromArchiveName( $row->oi_name, $row->oi_archive_name );
                    $checkPaths[] = $file->getPath();
                }

                foreach ( array_chunk( $checkPaths, $this->mBatchSize ) as $paths ) {
                    $be->preloadFileStat( [ 'srcs' => $paths ] );
                    foreach ( $paths as $path ) {
                        if ( $be->fileExists( [ 'src' => $path ] ) === false ) {
                            $this->output( "$path\n" );
                        }
                    }
                }
            }
        } while ( $res->numRows() >= $this->mBatchSize );
    }
}

$maintClass = 'FindMissingFiles';
require_once RUN_MAINTENANCE_IF_MAIN;