FileBackend.php

Go to the documentation of this file.

<?php
abstract class FileBackend {
        protected $name; // string; unique backend name
        protected $wikiId; // string; unique wiki name
        protected $readOnly; // string; read-only explanation message
        protected $lockManager;

        public function __construct( array $config ) {
                $this->name = $config['name'];
                if ( !preg_match( '!^[a-zA-Z0-9-_]{1,255}$!', $this->name ) ) {
                        throw new MWException( "Backend name `{$this->name}` is invalid." );
                }
                $this->wikiId = isset( $config['wikiId'] )
                        ? $config['wikiId']
                        : wfWikiID(); // e.g. "my_wiki-en_"
                $this->lockManager = ( $config['lockManager'] instanceof LockManager )
                        ? $config['lockManager']
                        : LockManagerGroup::singleton()->get( $config['lockManager'] );
                $this->readOnly = isset( $config['readOnly'] )
                        ? (string)$config['readOnly']
                        : '';
        }

        final public function getName() {
                return $this->name;
        }

        final public function isReadOnly() {
                return ( $this->readOnly != '' );
        }

        final public function getReadOnlyReason() {
                return ( $this->readOnly != '' ) ? $this->readOnly : false;
        }

        final public function doOperations( array $ops, array $opts = array() ) {
                if ( $this->isReadOnly() ) {
                        return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
                }
                if ( empty( $opts['force'] ) ) { // sanity
                        unset( $opts['nonLocking'] );
                        unset( $opts['allowStale'] );
                }
                return $this->doOperationsInternal( $ops, $opts );
        }

        abstract protected function doOperationsInternal( array $ops, array $opts );

        final public function doOperation( array $op, array $opts = array() ) {
                return $this->doOperations( array( $op ), $opts );
        }

        final public function create( array $params, array $opts = array() ) {
                $params['op'] = 'create';
                return $this->doOperation( $params, $opts );
        }

        final public function store( array $params, array $opts = array() ) {
                $params['op'] = 'store';
                return $this->doOperation( $params, $opts );
        }

        final public function copy( array $params, array $opts = array() ) {
                $params['op'] = 'copy';
                return $this->doOperation( $params, $opts );
        }

        final public function move( array $params, array $opts = array() ) {
                $params['op'] = 'move';
                return $this->doOperation( $params, $opts );
        }

        final public function delete( array $params, array $opts = array() ) {
                $params['op'] = 'delete';
                return $this->doOperation( $params, $opts );
        }

        abstract public function concatenate( array $params );

        final public function prepare( array $params ) {
                if ( $this->isReadOnly() ) {
                        return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
                }
                return $this->doPrepare( $params );
        }

        abstract protected function doPrepare( array $params );

        final public function secure( array $params ) {
                if ( $this->isReadOnly() ) {
                        return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
                }
                $status = $this->doPrepare( $params ); // dir must exist to restrict it
                if ( $status->isOK() ) {
                        $status->merge( $this->doSecure( $params ) );
                }
                return $status;
        }

        abstract protected function doSecure( array $params );

        final public function clean( array $params ) {
                if ( $this->isReadOnly() ) {
                        return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
                }
                return $this->doClean( $params );
        }

        abstract protected function doClean( array $params );

        abstract public function fileExists( array $params );

        abstract public function getFileTimestamp( array $params );

        abstract public function getFileContents( array $params );

        abstract public function getFileSize( array $params );

        abstract public function getFileStat( array $params );

        abstract public function getFileSha1Base36( array $params );

        abstract public function getFileProps( array $params );

        abstract public function streamFile( array $params );

        abstract public function getLocalReference( array $params );

        abstract public function getLocalCopy( array $params );

        abstract public function getFileList( array $params );

        public function clearCache( array $paths = null ) {}

        final public function lockFiles( array $paths, $type ) {
                return $this->lockManager->lock( $paths, $type );
        }

        final public function unlockFiles( array $paths, $type ) {
                return $this->lockManager->unlock( $paths, $type );
        }

        final public function getScopedFileLocks( array $paths, $type, Status $status ) {
                return ScopedLock::factory( $this->lockManager, $paths, $type, $status );
        }

        final public static function isStoragePath( $path ) {
                return ( strpos( $path, 'mwstore://' ) === 0 );
        }

        final public static function splitStoragePath( $storagePath ) {
                if ( self::isStoragePath( $storagePath ) ) {
                        // Remove the "mwstore://" prefix and split the path
                        $parts = explode( '/', substr( $storagePath, 10 ), 3 );
                        if ( count( $parts ) >= 2 && $parts[0] != '' && $parts[1] != '' ) {
                                if ( count( $parts ) == 3 ) {
                                        return $parts; // e.g. "backend/container/path"
                                } else {
                                        return array( $parts[0], $parts[1], '' ); // e.g. "backend/container" 
                                }
                        }
                }
                return array( null, null, null );
        }

        final public static function normalizeStoragePath( $storagePath ) {
                list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
                if ( $relPath !== null ) { // must be for this backend
                        $relPath = self::normalizeContainerPath( $relPath );
                        if ( $relPath !== null ) {
                                return ( $relPath != '' )
                                        ? "mwstore://{$backend}/{$container}/{$relPath}"
                                        : "mwstore://{$backend}/{$container}";
                        }
                }
                return null;
        }

        final protected static function normalizeContainerPath( $path ) {
                // Normalize directory separators
                $path = strtr( $path, '\\', '/' );
                // Collapse any consecutive directory separators
                $path = preg_replace( '![/]{2,}!', '/', $path );
                // Remove any leading directory separator
                $path = ltrim( $path, '/' );
                // Use the same traversal protection as Title::secureAndSplit()
                if ( strpos( $path, '.' ) !== false ) {
                        if (
                                $path === '.' ||
                                $path === '..' ||
                                strpos( $path, './' ) === 0 ||
                                strpos( $path, '../' ) === 0 ||
                                strpos( $path, '/./' ) !== false ||
                                strpos( $path, '/../' ) !== false
                        ) { 
                                return null;
                        }
                }
                return $path;
        }

        final public static function parentStoragePath( $storagePath ) {
                $storagePath = dirname( $storagePath );
                list( $b, $cont, $rel ) = self::splitStoragePath( $storagePath );
                return ( $rel === null ) ? null : $storagePath;
        }

        final public static function extensionFromPath( $path ) {
                $i = strrpos( $path, '.' );
                return strtolower( $i ? substr( $path, $i + 1 ) : '' );
        }
}

abstract class FileBackendStore extends FileBackend {
        protected $cache = array(); // (storage path => key => value)
        protected $maxCacheSize = 100; // integer; max paths with entries
        protected $expensiveCache = array(); // (storage path => key => value)
        protected $maxExpensiveCacheSize = 10; // integer; max paths with entries

        protected $shardViaHashLevels = array(); // (container name => config array)

        protected $maxFileSize = 1000000000; // integer bytes (1GB)

        final public function maxFileSizeInternal() {
                return $this->maxFileSize;
        }

        abstract public function isPathUsableInternal( $storagePath );

        final public function createInternal( array $params ) {
                wfProfileIn( __METHOD__ );
                if ( strlen( $params['content'] ) > $this->maxFileSizeInternal() ) {
                        $status = Status::newFatal( 'backend-fail-create', $params['dst'] );
                } else {
                        $status = $this->doCreateInternal( $params );
                        $this->clearCache( array( $params['dst'] ) );
                }
                wfProfileOut( __METHOD__ );
                return $status;
        }

        abstract protected function doCreateInternal( array $params );

        final public function storeInternal( array $params ) {
                wfProfileIn( __METHOD__ );
                if ( filesize( $params['src'] ) > $this->maxFileSizeInternal() ) {
                        $status = Status::newFatal( 'backend-fail-store', $params['dst'] );
                } else {
                        $status = $this->doStoreInternal( $params );
                        $this->clearCache( array( $params['dst'] ) );
                }
                wfProfileOut( __METHOD__ );
                return $status;
        }

        abstract protected function doStoreInternal( array $params );

        final public function copyInternal( array $params ) {
                wfProfileIn( __METHOD__ );
                $status = $this->doCopyInternal( $params );
                $this->clearCache( array( $params['dst'] ) );
                wfProfileOut( __METHOD__ );
                return $status;
        }

        abstract protected function doCopyInternal( array $params );

        final public function deleteInternal( array $params ) {
                wfProfileIn( __METHOD__ );
                $status = $this->doDeleteInternal( $params );
                $this->clearCache( array( $params['src'] ) );
                wfProfileOut( __METHOD__ );
                return $status;
        }

        abstract protected function doDeleteInternal( array $params );

        final public function moveInternal( array $params ) {
                wfProfileIn( __METHOD__ );
                $status = $this->doMoveInternal( $params );
                $this->clearCache( array( $params['src'], $params['dst'] ) );
                wfProfileOut( __METHOD__ );
                return $status;
        }

        protected function doMoveInternal( array $params ) {
                // Copy source to dest
                $status = $this->copyInternal( $params );
                if ( $status->isOK() ) {
                        // Delete source (only fails due to races or medium going down)
                        $status->merge( $this->deleteInternal( array( 'src' => $params['src'] ) ) );
                        $status->setResult( true, $status->value ); // ignore delete() errors
                }
                return $status;
        }

        final public function concatenate( array $params ) {
                wfProfileIn( __METHOD__ );
                $status = Status::newGood();

                // Try to lock the source files for the scope of this function
                $scopeLockS = $this->getScopedFileLocks( $params['srcs'], LockManager::LOCK_UW, $status );
                if ( $status->isOK() ) {
                        // Actually do the concatenation
                        $status->merge( $this->doConcatenate( $params ) );
                }

                wfProfileOut( __METHOD__ );
                return $status;
        }

        protected function doConcatenate( array $params ) {
                $status = Status::newGood();
                $tmpPath = $params['dst']; // convenience

                // Check that the specified temp file is valid...
                wfSuppressWarnings();
                $ok = ( is_file( $tmpPath ) && !filesize( $tmpPath ) );
                wfRestoreWarnings();
                if ( !$ok ) { // not present or not empty
                        $status->fatal( 'backend-fail-opentemp', $tmpPath );
                        return $status;
                }

                // Build up the temp file using the source chunks (in order)...
                $tmpHandle = fopen( $tmpPath, 'ab' );
                if ( $tmpHandle === false ) {
                        $status->fatal( 'backend-fail-opentemp', $tmpPath );
                        return $status;
                }
                foreach ( $params['srcs'] as $virtualSource ) {
                        // Get a local FS version of the chunk
                        $tmpFile = $this->getLocalReference( array( 'src' => $virtualSource ) );
                        if ( !$tmpFile ) {
                                $status->fatal( 'backend-fail-read', $virtualSource );
                                return $status;
                        }
                        // Get a handle to the local FS version
                        $sourceHandle = fopen( $tmpFile->getPath(), 'r' );
                        if ( $sourceHandle === false ) {
                                fclose( $tmpHandle );
                                $status->fatal( 'backend-fail-read', $virtualSource );
                                return $status;
                        }
                        // Append chunk to file (pass chunk size to avoid magic quotes)
                        if ( !stream_copy_to_stream( $sourceHandle, $tmpHandle ) ) {
                                fclose( $sourceHandle );
                                fclose( $tmpHandle );
                                $status->fatal( 'backend-fail-writetemp', $tmpPath );
                                return $status;
                        }
                        fclose( $sourceHandle );
                }
                if ( !fclose( $tmpHandle ) ) {
                        $status->fatal( 'backend-fail-closetemp', $tmpPath );
                        return $status;
                }

                clearstatcache(); // temp file changed

                return $status;
        }

        final protected function doPrepare( array $params ) {
                wfProfileIn( __METHOD__ );

                $status = Status::newGood();
                list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
                if ( $dir === null ) {
                        $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
                        wfProfileOut( __METHOD__ );
                        return $status; // invalid storage path
                }

                if ( $shard !== null ) { // confined to a single container/shard
                        $status->merge( $this->doPrepareInternal( $fullCont, $dir, $params ) );
                } else { // directory is on several shards
                        wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
                        list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
                        foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
                                $status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
                        }
                }

                wfProfileOut( __METHOD__ );
                return $status;
        }

        protected function doPrepareInternal( $container, $dir, array $params ) {
                return Status::newGood();
        }

        final protected function doSecure( array $params ) {
                wfProfileIn( __METHOD__ );
                $status = Status::newGood();

                list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
                if ( $dir === null ) {
                        $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
                        wfProfileOut( __METHOD__ );
                        return $status; // invalid storage path
                }

                if ( $shard !== null ) { // confined to a single container/shard
                        $status->merge( $this->doSecureInternal( $fullCont, $dir, $params ) );
                } else { // directory is on several shards
                        wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
                        list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
                        foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
                                $status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
                        }
                }

                wfProfileOut( __METHOD__ );
                return $status;
        }

        protected function doSecureInternal( $container, $dir, array $params ) {
                return Status::newGood();
        }

        final protected function doClean( array $params ) {
                wfProfileIn( __METHOD__ );
                $status = Status::newGood();

                list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
                if ( $dir === null ) {
                        $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
                        wfProfileOut( __METHOD__ );
                        return $status; // invalid storage path
                }

                // Attempt to lock this directory...
                $filesLockEx = array( $params['dir'] );
                $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
                if ( !$status->isOK() ) {
                        wfProfileOut( __METHOD__ );
                        return $status; // abort
                }

                if ( $shard !== null ) { // confined to a single container/shard
                        $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
                } else { // directory is on several shards
                        wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
                        list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
                        foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
                                $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
                        }
                }

                wfProfileOut( __METHOD__ );
                return $status;
        }

        protected function doCleanInternal( $container, $dir, array $params ) {
                return Status::newGood();
        }

        final public function fileExists( array $params ) {
                wfProfileIn( __METHOD__ );
                $stat = $this->getFileStat( $params );
                wfProfileOut( __METHOD__ );
                return ( $stat === null ) ? null : (bool)$stat; // null => failure
        }

        final public function getFileTimestamp( array $params ) {
                wfProfileIn( __METHOD__ );
                $stat = $this->getFileStat( $params );
                wfProfileOut( __METHOD__ );
                return $stat ? $stat['mtime'] : false;
        }

        final public function getFileSize( array $params ) {
                wfProfileIn( __METHOD__ );
                $stat = $this->getFileStat( $params );
                wfProfileOut( __METHOD__ );
                return $stat ? $stat['size'] : false;
        }

        final public function getFileStat( array $params ) {
                wfProfileIn( __METHOD__ );
                $path = self::normalizeStoragePath( $params['src'] );
                if ( $path === null ) {
                        return false; // invalid storage path
                }
                $latest = !empty( $params['latest'] );
                if ( isset( $this->cache[$path]['stat'] ) ) {
                        // If we want the latest data, check that this cached
                        // value was in fact fetched with the latest available data.
                        if ( !$latest || $this->cache[$path]['stat']['latest'] ) {
                                wfProfileOut( __METHOD__ );
                                return $this->cache[$path]['stat'];
                        }
                }
                $stat = $this->doGetFileStat( $params );
                if ( is_array( $stat ) ) { // don't cache negatives
                        $this->trimCache(); // limit memory
                        $this->cache[$path]['stat'] = $stat;
                        $this->cache[$path]['stat']['latest'] = $latest;
                }
                wfProfileOut( __METHOD__ );
                return $stat;
        }

        abstract protected function doGetFileStat( array $params );

        public function getFileContents( array $params ) {
                wfProfileIn( __METHOD__ );
                $tmpFile = $this->getLocalReference( $params );
                if ( !$tmpFile ) {
                        wfProfileOut( __METHOD__ );
                        return false;
                }
                wfSuppressWarnings();
                $data = file_get_contents( $tmpFile->getPath() );
                wfRestoreWarnings();
                wfProfileOut( __METHOD__ );
                return $data;
        }

        final public function getFileSha1Base36( array $params ) {
                wfProfileIn( __METHOD__ );
                $path = $params['src'];
                if ( isset( $this->cache[$path]['sha1'] ) ) {
                        wfProfileOut( __METHOD__ );
                        return $this->cache[$path]['sha1'];
                }
                $hash = $this->doGetFileSha1Base36( $params );
                if ( $hash ) { // don't cache negatives
                        $this->trimCache(); // limit memory
                        $this->cache[$path]['sha1'] = $hash;
                }
                wfProfileOut( __METHOD__ );
                return $hash;
        }

        protected function doGetFileSha1Base36( array $params ) {
                $fsFile = $this->getLocalReference( $params );
                if ( !$fsFile ) {
                        return false;
                } else {
                        return $fsFile->getSha1Base36();
                }
        }

        final public function getFileProps( array $params ) {
                wfProfileIn( __METHOD__ );
                $fsFile = $this->getLocalReference( $params );
                $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
                wfProfileOut( __METHOD__ );
                return $props;
        }

        public function getLocalReference( array $params ) {
                wfProfileIn( __METHOD__ );
                $path = $params['src'];
                if ( isset( $this->expensiveCache[$path]['localRef'] ) ) {
                        wfProfileOut( __METHOD__ );
                        return $this->expensiveCache[$path]['localRef'];
                }
                $tmpFile = $this->getLocalCopy( $params );
                if ( $tmpFile ) { // don't cache negatives
                        $this->trimExpensiveCache(); // limit memory
                        $this->expensiveCache[$path]['localRef'] = $tmpFile;
                }
                wfProfileOut( __METHOD__ );
                return $tmpFile;
        }

        final public function streamFile( array $params ) {
                wfProfileIn( __METHOD__ );
                $status = Status::newGood();

                $info = $this->getFileStat( $params );
                if ( !$info ) { // let StreamFile handle the 404
                        $status->fatal( 'backend-fail-notexists', $params['src'] );
                }

                // Set output buffer and HTTP headers for stream
                $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
                $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
                if ( $res == StreamFile::NOT_MODIFIED ) {
                        // do nothing; client cache is up to date
                } elseif ( $res == StreamFile::READY_STREAM ) {
                        $status = $this->doStreamFile( $params );
                } else {
                        $status->fatal( 'backend-fail-stream', $params['src'] );
                }

                wfProfileOut( __METHOD__ );
                return $status;
        }

        protected function doStreamFile( array $params ) {
                $status = Status::newGood();

                $fsFile = $this->getLocalReference( $params );
                if ( !$fsFile ) {
                        $status->fatal( 'backend-fail-stream', $params['src'] );
                } elseif ( !readfile( $fsFile->getPath() ) ) {
                        $status->fatal( 'backend-fail-stream', $params['src'] );
                }

                return $status;
        }

        final public function getFileList( array $params ) {
                list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
                if ( $dir === null ) { // invalid storage path
                        return null;
                }
                if ( $shard !== null ) {
                        // File listing is confined to a single container/shard
                        return $this->getFileListInternal( $fullCont, $dir, $params );
                } else {
                        wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
                        // File listing spans multiple containers/shards
                        list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
                        return new FileBackendStoreShardListIterator( $this,
                                $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
                }
        }

        abstract public function getFileListInternal( $container, $dir, array $params );

        protected function supportedOperations() {
                return array(
                        'store'       => 'StoreFileOp',
                        'copy'        => 'CopyFileOp',
                        'move'        => 'MoveFileOp',
                        'delete'      => 'DeleteFileOp',
                        'create'      => 'CreateFileOp',
                        'null'        => 'NullFileOp'
                );
        }

        final public function getOperations( array $ops ) {
                $supportedOps = $this->supportedOperations();

                $performOps = array(); // array of FileOp objects
                // Build up ordered array of FileOps...
                foreach ( $ops as $operation ) {
                        $opName = $operation['op'];
                        if ( isset( $supportedOps[$opName] ) ) {
                                $class = $supportedOps[$opName];
                                // Get params for this operation
                                $params = $operation;
                                // Append the FileOp class
                                $performOps[] = new $class( $this, $params );
                        } else {
                                throw new MWException( "Operation `$opName` is not supported." );
                        }
                }

                return $performOps;
        }

        protected function doOperationsInternal( array $ops, array $opts ) {
                wfProfileIn( __METHOD__ );
                $status = Status::newGood();

                // Build up a list of FileOps...
                $performOps = $this->getOperations( $ops );

                // Acquire any locks as needed...
                if ( empty( $opts['nonLocking'] ) ) {
                        // Build up a list of files to lock...
                        $filesLockEx = $filesLockSh = array();
                        foreach ( $performOps as $fileOp ) {
                                $filesLockSh = array_merge( $filesLockSh, $fileOp->storagePathsRead() );
                                $filesLockEx = array_merge( $filesLockEx, $fileOp->storagePathsChanged() );
                        }
                        // Optimization: if doing an EX lock anyway, don't also set an SH one
                        $filesLockSh = array_diff( $filesLockSh, $filesLockEx );
                        // Get a shared lock on the parent directory of each path changed
                        $filesLockSh = array_merge( $filesLockSh, array_map( 'dirname', $filesLockEx ) );
                        // Try to lock those files for the scope of this function...
                        $scopeLockS = $this->getScopedFileLocks( $filesLockSh, LockManager::LOCK_UW, $status );
                        $scopeLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
                        if ( !$status->isOK() ) {
                                wfProfileOut( __METHOD__ );
                                return $status; // abort
                        }
                }

                // Clear any cache entries (after locks acquired)
                $this->clearCache();

                // Actually attempt the operation batch...
                $subStatus = FileOp::attemptBatch( $performOps, $opts );

                // Merge errors into status fields
                $status->merge( $subStatus );
                $status->success = $subStatus->success; // not done in merge()

                wfProfileOut( __METHOD__ );
                return $status;
        }

        final public function clearCache( array $paths = null ) {
                if ( is_array( $paths ) ) {
                        $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
                        $paths = array_filter( $paths, 'strlen' ); // remove nulls
                }
                if ( $paths === null ) {
                        $this->cache = array();
                        $this->expensiveCache = array();
                } else {
                        foreach ( $paths as $path ) {
                                unset( $this->cache[$path] );
                                unset( $this->expensiveCache[$path] );
                        }
                }
                $this->doClearCache( $paths );
        }

        protected function doClearCache( array $paths = null ) {}

        protected function trimCache() {
                if ( count( $this->cache ) >= $this->maxCacheSize ) {
                        reset( $this->cache );
                        unset( $this->cache[key( $this->cache )] );
                }
        }

        protected function trimExpensiveCache() {
                if ( count( $this->expensiveCache ) >= $this->maxExpensiveCacheSize ) {
                        reset( $this->expensiveCache );
                        unset( $this->expensiveCache[key( $this->expensiveCache )] );
                }
        }

        final protected static function isValidContainerName( $container ) {
                // This accounts for Swift and S3 restrictions while leaving room 
                // for things like '.xxx' (hex shard chars) or '.seg' (segments).
                // This disallows directory separators or traversal characters.
                // Note that matching strings URL encode to the same string;
                // in Swift, the length restriction is *after* URL encoding.
                return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
        }

        final protected function resolveStoragePath( $storagePath ) {
                list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
                if ( $backend === $this->name ) { // must be for this backend
                        $relPath = self::normalizeContainerPath( $relPath );
                        if ( $relPath !== null ) {
                                // Get shard for the normalized path if this container is sharded
                                $cShard = $this->getContainerShard( $container, $relPath );
                                // Validate and sanitize the relative path (backend-specific)
                                $relPath = $this->resolveContainerPath( $container, $relPath );
                                if ( $relPath !== null ) {
                                        // Prepend any wiki ID prefix to the container name
                                        $container = $this->fullContainerName( $container );
                                        if ( self::isValidContainerName( $container ) ) {
                                                // Validate and sanitize the container name (backend-specific)
                                                $container = $this->resolveContainerName( "{$container}{$cShard}" );
                                                if ( $container !== null ) {
                                                        return array( $container, $relPath, $cShard );
                                                }
                                        }
                                }
                        }
                }
                return array( null, null, null );
        }

        final protected function resolveStoragePathReal( $storagePath ) {
                list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
                if ( $cShard !== null ) {
                        return array( $container, $relPath );
                }
                return array( null, null );
        }

        final protected function getContainerShard( $container, $relPath ) {
                list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
                if ( $levels == 1 || $levels == 2 ) {
                        // Hash characters are either base 16 or 36
                        $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
                        // Get a regex that represents the shard portion of paths.
                        // The concatenation of the captures gives us the shard.
                        if ( $levels === 1 ) { // 16 or 36 shards per container
                                $hashDirRegex = '(' . $char . ')';
                        } else { // 256 or 1296 shards per container
                                if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
                                        $hashDirRegex = $char . '/(' . $char . '{2})';
                                } else { // short hash dir format (e.g. "a/b/c")
                                        $hashDirRegex = '(' . $char . ')/(' . $char . ')';
                                }
                        }
                        // Allow certain directories to be above the hash dirs so as
                        // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
                        // They must be 2+ chars to avoid any hash directory ambiguity.
                        $m = array();
                        if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
                                return '.' . implode( '', array_slice( $m, 1 ) );
                        }
                        return null; // failed to match
                }
                return ''; // no sharding
        }

        final protected function getContainerHashLevels( $container ) {
                if ( isset( $this->shardViaHashLevels[$container] ) ) {
                        $config = $this->shardViaHashLevels[$container];
                        $hashLevels = (int)$config['levels'];
                        if ( $hashLevels == 1 || $hashLevels == 2 ) {
                                $hashBase = (int)$config['base'];
                                if ( $hashBase == 16 || $hashBase == 36 ) {
                                        return array( $hashLevels, $hashBase, $config['repeat'] );
                                }
                        }
                }
                return array( 0, 0, false ); // no sharding
        }

        final protected function getContainerSuffixes( $container ) {
                $shards = array();
                list( $digits, $base ) = $this->getContainerHashLevels( $container );
                if ( $digits > 0 ) {
                        $numShards = pow( $base, $digits );
                        for ( $index = 0; $index < $numShards; $index++ ) {
                                $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
                        }
                }
                return $shards;
        }

        final protected function fullContainerName( $container ) {
                if ( $this->wikiId != '' ) {
                        return "{$this->wikiId}-$container";
                } else {
                        return $container;
                }
        }

        protected function resolveContainerName( $container ) {
                return $container;
        }

        protected function resolveContainerPath( $container, $relStoragePath ) {
                return $relStoragePath;
        }
}

class FileBackendStoreShardListIterator implements Iterator {
        /* @var FileBackendStore */
        protected $backend;
        /* @var Array */
        protected $params;
        /* @var Array */
        protected $shardSuffixes;
        protected $container; // string
        protected $directory; // string

        /* @var Traversable */
        protected $iter;
        protected $curShard = 0; // integer
        protected $pos = 0; // integer

        public function __construct(
                FileBackendStore $backend, $container, $dir, array $suffixes, array $params
        ) {
                $this->backend = $backend;
                $this->container = $container;
                $this->directory = $dir;
                $this->shardSuffixes = $suffixes;
                $this->params = $params;
        }

        public function current() {
                if ( is_array( $this->iter ) ) {
                        return current( $this->iter );
                } else {
                        return $this->iter->current();
                }
        }

        public function key() {
                return $this->pos;
        }

        public function next() {
                ++$this->pos;
                if ( is_array( $this->iter ) ) {
                        next( $this->iter );
                } else {
                        $this->iter->next();
                }
                // Find the next non-empty shard if no elements are left
                $this->nextShardIteratorIfNotValid();
        }

        protected function nextShardIteratorIfNotValid() {
                while ( !$this->valid() ) {
                        if ( ++$this->curShard >= count( $this->shardSuffixes ) ) {
                                break; // no more container shards
                        }
                        $this->setIteratorFromCurrentShard();
                }
        }

        protected function setIteratorFromCurrentShard() {
                $suffix = $this->shardSuffixes[$this->curShard];
                $this->iter = $this->backend->getFileListInternal(
                        "{$this->container}{$suffix}", $this->directory, $this->params );
        }

        public function rewind() {
                $this->pos = 0;
                $this->curShard = 0;
                $this->setIteratorFromCurrentShard();
                // Find the next non-empty shard if this one has no elements
                $this->nextShardIteratorIfNotValid();
        }

        public function valid() {
                if ( $this->iter == null ) {
                        return false; // some failure?
                } elseif ( is_array( $this->iter ) ) {
                        return ( current( $this->iter ) !== false ); // no paths can have this value
                } else {
                        return $this->iter->valid();
                }
        }
}