cleanupUploadStash.php

Go to the documentation of this file.

<?php
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
 
require_once __DIR__ . '/Maintenance.php';
 
class CleanupUploadStash extends Maintenance {
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Clean up abandoned files in temporary uploaded file stash' );
        $this->setBatchSize( 50 );
    }
 
    public function execute() {
        $repo = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo();
        $tempRepo = $repo->getTempRepo();
 
        $dbr = $repo->getReplicaDB();
 
        // how far back should this look for files to delete?
        $cutoff = time() - (int)$this->getConfig()->get( MainConfigNames::UploadStashMaxAge );
 
        $this->output( "Getting list of files to clean up...\n" );
        $res = $dbr->newSelectQueryBuilder()
            ->select( 'us_key' )
            ->from( 'uploadstash' )
            ->where( 'us_timestamp < ' . $dbr->addQuotes( $dbr->timestamp( $cutoff ) ) )
            ->caller( __METHOD__ )
            ->fetchResultSet();
 
        // Delete all registered stash files...
        if ( $res->numRows() == 0 ) {
            $this->output( "No stashed files to cleanup according to the DB.\n" );
        } else {
            // finish the read before starting writes.
            $keys = [];
            foreach ( $res as $row ) {
                array_push( $keys, $row->us_key );
            }
 
            $this->output( 'Removing ' . count( $keys ) . " file(s)...\n" );
            // this could be done some other, more direct/efficient way, but using
            // UploadStash's own methods means it's less likely to fall accidentally
            // out-of-date someday
            $stash = new UploadStash( $repo );
 
            $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 
            $i = 0;
            foreach ( $keys as $key ) {
                $i++;
                try {
                    $stash->getFile( $key, true );
                    $stash->removeFileNoAuth( $key );
                } catch ( UploadStashException $ex ) {
                    $type = get_class( $ex );
                    $this->output( "Failed removing stashed upload with key: $key ($type)\n" );
                }
                if ( $i % 100 == 0 ) {
                    $lbFactory->waitForReplication();
                    $this->output( "$i\n" );
                }
            }
            $this->output( "$i done\n" );
        }
 
        // Delete all the corresponding thumbnails...
        $dir = $tempRepo->getZonePath( 'thumb' );
        $iterator = $tempRepo->getBackend()->getFileList( [ 'dir' => $dir, 'adviseStat' => 1 ] );
        if ( $iterator === null ) {
            $this->fatalError( "Could not get file listing." );
        }
        $this->output( "Deleting old thumbnails...\n" );
        $i = 0;
        $batch = [];
        foreach ( $iterator as $file ) {
            if ( wfTimestamp( TS_UNIX, $tempRepo->getFileTimestamp( "$dir/$file" ) ) < $cutoff ) {
                $batch[] = [ 'op' => 'delete', 'src' => "$dir/$file" ];
                if ( count( $batch ) >= $this->getBatchSize() ) {
                    $this->doOperations( $tempRepo, $batch );
                    $i += count( $batch );
                    $batch = [];
                    $this->output( "$i\n" );
                }
            }
        }
        if ( count( $batch ) ) {
            $this->doOperations( $tempRepo, $batch );
            $i += count( $batch );
        }
        $this->output( "$i done\n" );
 
        // Apparently lots of stash files are not registered in the DB...
        $dir = $tempRepo->getZonePath( 'public' );
        $iterator = $tempRepo->getBackend()->getFileList( [ 'dir' => $dir, 'adviseStat' => 1 ] );
        if ( $iterator === null ) {
            $this->fatalError( "Could not get file listing." );
        }
        $this->output( "Deleting orphaned temp files...\n" );
        if ( strpos( $dir, '/local-temp' ) === false ) {
            $this->fatalError( "Temp repo is not using the temp container." );
        }
 
        $i = 0;
        $batch = [];
        foreach ( $iterator as $file ) {
            if ( wfTimestamp( TS_UNIX, $tempRepo->getFileTimestamp( "$dir/$file" ) ) < $cutoff ) {
                $batch[] = [ 'op' => 'delete', 'src' => "$dir/$file" ];
                if ( count( $batch ) >= $this->getBatchSize() ) {
                    $this->doOperations( $tempRepo, $batch );
                    $i += count( $batch );
                    $batch = [];
                    $this->output( "$i\n" );
                }
            }
        }
        if ( count( $batch ) ) {
            $this->doOperations( $tempRepo, $batch );
            $i += count( $batch );
        }
        $this->output( "$i done\n" );
    }
 
    protected function doOperations( FileRepo $tempRepo, array $ops ) {
        $status = $tempRepo->getBackend()->doQuickOperations( $ops );
        if ( !$status->isOK() ) {
            $this->error( print_r( Status::wrap( $status )->getErrorsArray(), true ) );
        }
    }
}
 
$maintClass = CleanupUploadStash::class;
require_once RUN_MAINTENANCE_IF_MAIN;