<?php
+/**
+ * @defgroup FileBackend File backend
+ * @ingroup FileRepo
+ *
+ * File backend is used to interact with file storage systems,
+ * such as the local file system, NFS, or cloud storage systems.
+ */
+
/**
* @file
* @ingroup FileBackend
/**
* Base class for all file backend classes (including multi-write backends).
+ *
* This class defines the methods as abstract that subclasses must implement.
* Outside callers can assume that all backends will have these functions.
*
- * All "storage paths" are of the format "mwstore://backend/container/path".
- * The paths use UNIX file system (FS) notation, though any particular backend may
- * not actually be using a local filesystem. Therefore, the paths are only virtual.
+ * All "storage paths" are of the format "mwstore://<backend>/<container>/<path>".
+ * The <path> portion is a relative path that uses UNIX file system (FS) notation,
+ * though any particular backend may not actually be using a local filesystem.
+ * Therefore, the relative paths are only virtual.
*
* Backend contents are stored under wiki-specific container names by default.
* For legacy reasons, this has no effect for the FS backend class, and per-wiki
* @ingroup FileBackend
* @since 1.19
*/
-abstract class FileBackendBase {
- protected $name; // unique backend name
- protected $wikiId; // unique wiki name
- protected $readOnly; // string
+abstract class FileBackend {
+ protected $name; // string; unique backend name
+ protected $wikiId; // string; unique wiki name
+ protected $readOnly; // string; read-only explanation message
/** @var LockManager */
protected $lockManager;
*
* $config includes:
* 'name' : The unique name of this backend.
+ * This should consist of alphanumberic, '-', and '_' characters.
* 'wikiId' : Prefix to container names that is unique to this wiki.
- * This should consist of alphanumberic, '-', and '_' chars.
+ * This should consist of alphanumberic, '-', and '_' characters.
* 'lockManager' : Registered name of a file lock manager to use.
* 'readOnly' : Write operations are disallowed if this is a non-empty string.
* It should be an explanation for the backend being read-only.
$this->wikiId = isset( $config['wikiId'] )
? $config['wikiId']
: wfWikiID(); // e.g. "my_wiki-en_"
- $this->wikiId = $this->resolveWikiId( $this->wikiId );
- $this->lockManager = LockManagerGroup::singleton()->get( $config['lockManager'] );
+ $this->lockManager = ( $config['lockManager'] instanceof LockManager )
+ ? $config['lockManager']
+ : LockManagerGroup::singleton()->get( $config['lockManager'] );
$this->readOnly = isset( $config['readOnly'] )
? (string)$config['readOnly']
: '';
}
- /**
- * Normalize a wiki ID by replacing characters that are
- * not supported by the backend as part of container names.
- *
- * @param $wikiId string
- * @return string
- */
- protected function resolveWikiId( $wikiId ) {
- return $wikiId;
- }
-
/**
* Get the unique backend name.
* We may have multiple different backends of the same type.
return $this->name;
}
+ /**
+ * Check if this backend is read-only
+ *
+ * @return bool
+ */
+ final public function isReadOnly() {
+ return ( $this->readOnly != '' );
+ }
+
+ /**
+ * Get an explanatory message if this backend is read-only
+ *
+ * @return string|bool Returns falls if the backend is not read-only
+ */
+ final public function getReadOnlyReason() {
+ return ( $this->readOnly != '' ) ? $this->readOnly : false;
+ }
+
/**
* This is the main entry point into the backend for write operations.
* Callers supply an ordered list of operations to perform as a transaction.
+ * Files will be locked, the stat cache cleared, and then the operations attempted.
* If any serious errors occur, all attempted operations will be rolled back.
*
* $ops is an array of arrays. The outer array holds a list of operations.
* This can increase performance for non-critical writes.
* This has no effect unless the 'force' flag is set.
*
- * Remarks:
+ * Remarks on locking:
* File system paths given to operations should refer to files that are
- * either locked or otherwise safe from modification from other processes.
+ * already locked or otherwise safe from modification from other processes.
* Normally these files will be new temp files, which should be adequate.
*
* Return value:
* @return Status
*/
final public function doOperations( array $ops, array $opts = array() ) {
- if ( $this->readOnly != '' ) {
+ if ( $this->isReadOnly() ) {
return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
}
if ( empty( $opts['force'] ) ) { // sanity
}
/**
- * @see FileBackendBase::doOperations()
+ * @see FileBackend::doOperations()
*/
abstract protected function doOperationsInternal( array $ops, array $opts );
* If you are doing a batch of operations that should either
* all succeed or all fail, then use that function instead.
*
- * @see FileBackendBase::doOperations()
+ * @see FileBackend::doOperations()
*
* @param $op Array Operation
* @param $opts Array Operation options
* Performs a single create operation.
* This sets $params['op'] to 'create' and passes it to doOperation().
*
- * @see FileBackendBase::doOperation()
+ * @see FileBackend::doOperation()
*
* @param $params Array Operation parameters
* @param $opts Array Operation options
* Performs a single store operation.
* This sets $params['op'] to 'store' and passes it to doOperation().
*
- * @see FileBackendBase::doOperation()
+ * @see FileBackend::doOperation()
*
* @param $params Array Operation parameters
* @param $opts Array Operation options
* Performs a single copy operation.
* This sets $params['op'] to 'copy' and passes it to doOperation().
*
- * @see FileBackendBase::doOperation()
+ * @see FileBackend::doOperation()
*
* @param $params Array Operation parameters
* @param $opts Array Operation options
* Performs a single move operation.
* This sets $params['op'] to 'move' and passes it to doOperation().
*
- * @see FileBackendBase::doOperation()
+ * @see FileBackend::doOperation()
*
* @param $params Array Operation parameters
* @param $opts Array Operation options
* Performs a single delete operation.
* This sets $params['op'] to 'delete' and passes it to doOperation().
*
- * @see FileBackendBase::doOperation()
+ * @see FileBackend::doOperation()
*
* @param $params Array Operation parameters
* @param $opts Array Operation options
}
/**
- * Concatenate a list of storage files into a single file on the file system
+ * Concatenate a list of storage files into a single file system file.
+ * The target path should refer to a file that is already locked or
+ * otherwise safe from modification from other processes. Normally,
+ * the file will be a new temp file, which should be adequate.
* $params include:
* srcs : ordered source storage paths (e.g. chunk1, chunk2, ...)
* dst : file system path to 0-byte temp file
* @return Status
*/
final public function prepare( array $params ) {
- if ( $this->readOnly != '' ) {
+ if ( $this->isReadOnly() ) {
return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
}
return $this->doPrepare( $params );
}
/**
- * @see FileBackendBase::prepare()
+ * @see FileBackend::prepare()
*/
abstract protected function doPrepare( array $params );
* @return Status
*/
final public function secure( array $params ) {
- if ( $this->readOnly != '' ) {
+ if ( $this->isReadOnly() ) {
return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
}
$status = $this->doPrepare( $params ); // dir must exist to restrict it
}
/**
- * @see FileBackendBase::secure()
+ * @see FileBackend::secure()
*/
abstract protected function doSecure( array $params );
* @return Status
*/
final public function clean( array $params ) {
- if ( $this->readOnly != '' ) {
+ if ( $this->isReadOnly() ) {
return Status::newFatal( 'backend-fail-readonly', $this->name, $this->readOnly );
}
return $this->doClean( $params );
}
/**
- * @see FileBackendBase::clean()
+ * @see FileBackend::clean()
*/
abstract protected function doClean( array $params );
* latest : use the latest available data
*
* @param $params Array
- * @return string|false TS_MW timestamp or false on failure
+ * @return string|bool TS_MW timestamp or false on failure
*/
abstract public function getFileTimestamp( array $params );
* latest : use the latest available data
*
* @param $params Array
- * @return string|false Returns false on failure
+ * @return string|bool Returns false on failure
*/
abstract public function getFileContents( array $params );
* latest : use the latest available data
*
* @param $params Array
- * @return integer|false Returns false on failure
+ * @return integer|bool Returns false on failure
*/
abstract public function getFileSize( array $params );
* latest : use the latest available data
*
* @param $params Array
- * @return Array|false|null Returns null on failure
+ * @return Array|bool|null Returns null on failure
*/
abstract public function getFileStat( array $params );
* latest : use the latest available data
*
* @param $params Array
- * @return string|false Hash string or false on failure
+ * @return string|bool Hash string or false on failure
*/
abstract public function getFileSha1Base36( array $params );
* Write operations should *never* be done on this file as some backends
* may do internal tracking or may be instances of FileBackendMultiWrite.
* In that later case, there are copies of the file that must stay in sync.
+ * Additionally, further calls to this function may return the same file.
*
* $params include:
* src : source storage path
/**
* Get an iterator to list out all stored files under a storage directory.
- * If the directory is of the form "mwstore://container", then all items in
- * the container should be listed. If of the form "mwstore://container/dir",
- * then all items under that container directory should be listed.
+ * If the directory is of the form "mwstore://backend/container",
+ * then all files in the container should be listed.
+ * If the directory is of form "mwstore://backend/container/dir",
+ * then all files under that container directory should be listed.
* Results should be storage paths relative to the given directory.
*
+ * Storage backends with eventual consistency might return stale data.
+ *
* $params include:
* dir : storage path directory
*
* @param $paths Array Storage paths (optional)
* @return void
*/
- abstract public function clearCache( array $paths = null );
+ public function clearCache( array $paths = null ) {}
/**
* Lock the files at the given storage paths in the backend.
final public function getScopedFileLocks( array $paths, $type, Status $status ) {
return ScopedLock::factory( $this->lockManager, $paths, $type, $status );
}
+
+ /**
+ * Check if a given path is a "mwstore://" path.
+ * This does not do any further validation or any existence checks.
+ *
+ * @param $path string
+ * @return bool
+ */
+ final public static function isStoragePath( $path ) {
+ return ( strpos( $path, 'mwstore://' ) === 0 );
+ }
+
+ /**
+ * Split a storage path into a backend name, a container name,
+ * and a relative file path. The relative path may be the empty string.
+ * This does not do any path normalization or traversal checks.
+ *
+ * @param $storagePath string
+ * @return Array (backend, container, rel object) or (null, null, null)
+ */
+ 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 );
+ }
+
+ /**
+ * Normalize a storage path by cleaning up directory separators.
+ * Returns null if the path is not of the format of a valid storage path.
+ *
+ * @param $storagePath string
+ * @return string|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;
+ }
+
+ /**
+ * Validate and normalize a relative storage path.
+ * Null is returned if the path involves directory traversal.
+ * Traversal is insecure for FS backends and broken for others.
+ *
+ * @param $path string Storage path relative to a container
+ * @return string|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;
+ }
+
+ /**
+ * Get the parent storage directory of a storage path.
+ * This returns a path like "mwstore://backend/container",
+ * "mwstore://backend/container/...", or null if there is no parent.
+ *
+ * @param $storagePath string
+ * @return string|null
+ */
+ final public static function parentStoragePath( $storagePath ) {
+ $storagePath = dirname( $storagePath );
+ list( $b, $cont, $rel ) = self::splitStoragePath( $storagePath );
+ return ( $rel === null ) ? null : $storagePath;
+ }
+
+ /**
+ * Get the final extension from a storage or FS path
+ *
+ * @param $path string
+ * @return string
+ */
+ final public static function extensionFromPath( $path ) {
+ $i = strrpos( $path, '.' );
+ return strtolower( $i ? substr( $path, $i + 1 ) : '' );
+ }
}
/**
- * Base class for all single-write backends.
+ * @brief Base class for all backends using particular storage medium.
+ *
* This class defines the methods as abstract that subclasses must implement.
- * Callers outside of FileBackend and its helper classes, such as FileOp,
- * should only call functions that are present in FileBackendBase.
+ * Outside callers should *not* use functions with "Internal" in the name.
*
- * The FileBackendBase operations are implemented using primitive functions
+ * The FileBackend operations are implemented using basic functions
* such as storeInternal(), copyInternal(), deleteInternal() and the like.
* This class is also responsible for path resolution and sanitization.
*
* @ingroup FileBackend
* @since 1.19
*/
-abstract class FileBackend extends FileBackendBase {
- /** @var Array */
+abstract class FileBackendStore extends FileBackend {
+ /** @var Array Map of paths to small (RAM/disk) cache items */
protected $cache = array(); // (storage path => key => value)
- protected $maxCacheSize = 75; // integer; max paths with entries
- /** @var Array */
- protected $shardViaHashLevels = array(); // (container name => integer)
+ protected $maxCacheSize = 100; // integer; max paths with entries
+ /** @var Array Map of paths to large (RAM/disk) cache items */
+ protected $expensiveCache = array(); // (storage path => key => value)
+ protected $maxExpensiveCacheSize = 10; // integer; max paths with entries
+
+ /** @var Array Map of container names to sharding settings */
+ protected $shardViaHashLevels = array(); // (container name => config array)
protected $maxFileSize = 1000000000; // integer bytes (1GB)
* @return Status
*/
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;
}
/**
- * @see FileBackend::createInternal()
+ * @see FileBackendStore::createInternal()
*/
abstract protected function doCreateInternal( array $params );
* @return Status
*/
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;
}
/**
- * @see FileBackend::storeInternal()
+ * @see FileBackendStore::storeInternal()
*/
abstract protected function doStoreInternal( array $params );
* @return Status
*/
final public function copyInternal( array $params ) {
+ wfProfileIn( __METHOD__ );
$status = $this->doCopyInternal( $params );
$this->clearCache( array( $params['dst'] ) );
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::copyInternal()
+ * @see FileBackendStore::copyInternal()
*/
abstract protected function doCopyInternal( array $params );
* @return Status
*/
final public function deleteInternal( array $params ) {
+ wfProfileIn( __METHOD__ );
$status = $this->doDeleteInternal( $params );
$this->clearCache( array( $params['src'] ) );
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::deleteInternal()
+ * @see FileBackendStore::deleteInternal()
*/
abstract protected function doDeleteInternal( array $params );
* @return Status
*/
final public function moveInternal( array $params ) {
+ wfProfileIn( __METHOD__ );
$status = $this->doMoveInternal( $params );
$this->clearCache( array( $params['src'], $params['dst'] ) );
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::moveInternal()
+ * @see FileBackendStore::moveInternal()
+ * @return Status
*/
protected function doMoveInternal( array $params ) {
// Copy source to dest
$status = $this->copyInternal( $params );
- if ( !$status->isOK() ) {
- return $status;
+ 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
}
- // 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;
}
/**
- * @see FileBackendBase::concatenate()
+ * @see FileBackend::concatenate()
+ * @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() ) {
- return $status; // abort
+ if ( $status->isOK() ) {
+ // Actually do the concatenation
+ $status->merge( $this->doConcatenate( $params ) );
}
- // Actually do the concatenation
- $status->merge( $this->doConcatenate( $params ) );
-
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::concatenate()
+ * @see FileBackendStore::concatenate()
+ * @return Status
*/
protected function doConcatenate( array $params ) {
$status = Status::newGood();
}
// Build up the temp file using the source chunks (in order)...
- $tmpHandle = fopen( $tmpPath, 'a' );
+ $tmpHandle = fopen( $tmpPath, 'ab' );
if ( $tmpHandle === false ) {
$status->fatal( 'backend-fail-opentemp', $tmpPath );
return $status;
}
/**
- * @see FileBackendBase::doPrepare()
+ * @see FileBackend::doPrepare()
+ * @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
$status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
}
}
+
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::doPrepare()
+ * @see FileBackendStore::doPrepare()
+ * @return Status
*/
protected function doPrepareInternal( $container, $dir, array $params ) {
return Status::newGood();
}
/**
- * @see FileBackendBase::doSecure()
+ * @see FileBackend::doSecure()
+ * @return Status
*/
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
$status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
}
}
+
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::doSecure()
+ * @see FileBackendStore::doSecure()
+ * @return Status
*/
protected function doSecureInternal( $container, $dir, array $params ) {
return Status::newGood();
}
/**
- * @see FileBackendBase::doClean()
+ * @see FileBackend::doClean()
+ * @return Status
*/
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
$status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
}
}
+
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::doClean()
+ * @see FileBackendStore::doClean()
+ * @return Status
*/
protected function doCleanInternal( $container, $dir, array $params ) {
return Status::newGood();
}
/**
- * @see FileBackendBase::fileExists()
+ * @see FileBackend::fileExists()
+ * @return bool|null
*/
final public function fileExists( array $params ) {
+ wfProfileIn( __METHOD__ );
$stat = $this->getFileStat( $params );
- if ( $stat === null ) {
- return null; // failure
- }
- return (bool)$stat;
+ wfProfileOut( __METHOD__ );
+ return ( $stat === null ) ? null : (bool)$stat; // null => failure
}
/**
- * @see FileBackendBase::getFileTimestamp()
+ * @see FileBackend::getFileTimestamp()
+ * @return bool
*/
final public function getFileTimestamp( array $params ) {
+ wfProfileIn( __METHOD__ );
$stat = $this->getFileStat( $params );
- if ( $stat ) {
- return $stat['mtime'];
- } else {
- return false;
- }
+ wfProfileOut( __METHOD__ );
+ return $stat ? $stat['mtime'] : false;
}
/**
- * @see FileBackendBase::getFileSize()
+ * @see FileBackend::getFileSize()
+ * @return bool
*/
final public function getFileSize( array $params ) {
+ wfProfileIn( __METHOD__ );
$stat = $this->getFileStat( $params );
- if ( $stat ) {
- return $stat['size'];
- } else {
- return false;
- }
+ wfProfileOut( __METHOD__ );
+ return $stat ? $stat['size'] : false;
}
/**
- * @see FileBackendBase::getFileStat()
+ * @see FileBackend::getFileStat()
+ * @return bool
*/
final public function getFileStat( array $params ) {
- $path = $params['src'];
+ wfProfileIn( __METHOD__ );
+ $path = self::normalizeStoragePath( $params['src'] );
+ if ( $path === null ) {
+ wfProfileOut( __METHOD__ );
+ 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'];
}
}
+ wfProfileIn( __METHOD__ . '-miss' );
$stat = $this->doGetFileStat( $params );
+ wfProfileOut( __METHOD__ . '-miss' );
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;
}
/**
- * @see FileBackend::getFileStat()
+ * @see FileBackendStore::getFileStat()
*/
abstract protected function doGetFileStat( array $params );
/**
- * @see FileBackendBase::getFileContents()
+ * @see FileBackend::getFileContents()
+ * @return bool|string
*/
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;
}
/**
- * @see FileBackendBase::getFileSha1Base36()
+ * @see FileBackend::getFileSha1Base36()
+ * @return bool|string
*/
final public function getFileSha1Base36( array $params ) {
+ wfProfileIn( __METHOD__ );
$path = $params['src'];
if ( isset( $this->cache[$path]['sha1'] ) ) {
+ wfProfileOut( __METHOD__ );
return $this->cache[$path]['sha1'];
}
+ wfProfileIn( __METHOD__ . '-miss' );
$hash = $this->doGetFileSha1Base36( $params );
+ wfProfileOut( __METHOD__ . '-miss' );
if ( $hash ) { // don't cache negatives
$this->trimCache(); // limit memory
$this->cache[$path]['sha1'] = $hash;
}
+ wfProfileOut( __METHOD__ );
return $hash;
}
/**
- * @see FileBackend::getFileSha1Base36()
+ * @see FileBackendStore::getFileSha1Base36()
+ * @return bool
*/
protected function doGetFileSha1Base36( array $params ) {
$fsFile = $this->getLocalReference( $params );
}
/**
- * @see FileBackendBase::getFileProps()
+ * @see FileBackend::getFileProps()
+ * @return Array
*/
- public function getFileProps( array $params ) {
+ final public function getFileProps( array $params ) {
+ wfProfileIn( __METHOD__ );
$fsFile = $this->getLocalReference( $params );
- if ( !$fsFile ) {
- return FSFile::placeholderProps();
- } else {
- return $fsFile->getProps();
- }
+ $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
+ wfProfileOut( __METHOD__ );
+ return $props;
}
/**
- * @see FileBackendBase::getLocalReference()
+ * @see FileBackend::getLocalReference()
+ * @return TempFSFile|null
*/
public function getLocalReference( array $params ) {
+ wfProfileIn( __METHOD__ );
$path = $params['src'];
- if ( isset( $this->cache[$path]['localRef'] ) ) {
- return $this->cache[$path]['localRef'];
+ if ( isset( $this->expensiveCache[$path]['localRef'] ) ) {
+ wfProfileOut( __METHOD__ );
+ return $this->expensiveCache[$path]['localRef'];
}
$tmpFile = $this->getLocalCopy( $params );
if ( $tmpFile ) { // don't cache negatives
- $this->trimCache(); // limit memory
- $this->cache[$path]['localRef'] = $tmpFile;
+ $this->trimExpensiveCache(); // limit memory
+ $this->expensiveCache[$path]['localRef'] = $tmpFile;
}
+ wfProfileOut( __METHOD__ );
return $tmpFile;
}
/**
- * @see FileBackendBase::streamFile()
+ * @see FileBackend::streamFile()
+ * @return Status
*/
final public function streamFile( array $params ) {
+ wfProfileIn( __METHOD__ );
$status = Status::newGood();
$info = $this->getFileStat( $params );
$status->fatal( 'backend-fail-stream', $params['src'] );
}
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackend::streamFile()
+ * @see FileBackendStore::streamFile()
+ * @return Status
*/
protected function doStreamFile( array $params ) {
$status = Status::newGood();
}
/**
- * @see FileBackendBase::getFileList()
+ * @copydoc FileBackend::getFileList()
+ * @return Array|null|Traversable
*/
final public function getFileList( array $params ) {
list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
// File listing spans multiple containers/shards
list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
- return new FileBackendShardListIterator( $this,
- $fullCont, $this->getContainerSuffixes( $shortCont ), $params );
+ return new FileBackendStoreShardListIterator( $this,
+ $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
}
}
/**
- * Do not call this function from places outside FileBackend and ContainerFileListIterator
+ * Do not call this function from places outside FileBackend
*
+ * @see FileBackendStore::getFileList()
+ *
* @param $container string Resolved container name
* @param $dir string Resolved path relative to container
* @param $params Array
- * @see FileBackend::getFileList()
+ * @return Traversable|Array|null
*/
abstract public function getFileListInternal( $container, $dir, array $params );
}
/**
- * @see FileBackendBase::doOperationsInternal()
+ * @see FileBackend::doOperationsInternal()
+ * @return Status
*/
protected function doOperationsInternal( array $ops, array $opts ) {
+ wfProfileIn( __METHOD__ );
$status = Status::newGood();
// Build up a list of FileOps...
$scopeLockS = $this->getScopedFileLocks( $filesLockSh, LockManager::LOCK_UW, $status );
$scopeLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
if ( !$status->isOK() ) {
+ wfProfileOut( __METHOD__ );
return $status; // abort
}
}
$status->merge( $subStatus );
$status->success = $subStatus->success; // not done in merge()
+ wfProfileOut( __METHOD__ );
return $status;
}
/**
- * @see FileBackendBase::clearCache()
+ * @see FileBackend::clearCache()
*/
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 );
/**
* Clears any additional stat caches for storage paths
*
- * @see FileBackendBase::clearCache()
+ * @see FileBackend::clearCache()
*
* @param $paths Array Storage paths (optional)
* @return void
protected function doClearCache( array $paths = null ) {}
/**
- * Prune the cache if it is too big to add an item
+ * Prune the inexpensive cache if it is too big to add an item
*
* @return void
*/
protected function trimCache() {
if ( count( $this->cache ) >= $this->maxCacheSize ) {
reset( $this->cache );
- $key = key( $this->cache );
- unset( $this->cache[$key] );
+ unset( $this->cache[key( $this->cache )] );
}
}
/**
- * Check if a given path is a mwstore:// path.
- * This does not do any actual validation or existence checks.
+ * Prune the expensive cache if it is too big to add an item
*
- * @param $path string
- * @return bool
- */
- final public static function isStoragePath( $path ) {
- return ( strpos( $path, 'mwstore://' ) === 0 );
- }
-
- /**
- * Split a storage path (e.g. "mwstore://backend/container/path/to/object")
- * into a backend name, a container name, and a relative object path.
- *
- * @param $storagePath string
- * @return Array (backend, container, rel object) or (null, null, null)
+ * @return void
*/
- final public static function splitStoragePath( $storagePath ) {
- if ( self::isStoragePath( $storagePath ) ) {
- // Note: strlen( 'mwstore://' ) = 10
- $parts = explode( '/', substr( $storagePath, 10 ), 3 );
- if ( count( $parts ) == 3 ) {
- return $parts; // e.g. "backend/container/path"
- } elseif ( count( $parts ) == 2 ) {
- return array( $parts[0], $parts[1], '' ); // e.g. "backend/container"
- }
+ protected function trimExpensiveCache() {
+ if ( count( $this->expensiveCache ) >= $this->maxExpensiveCacheSize ) {
+ reset( $this->expensiveCache );
+ unset( $this->expensiveCache[key( $this->expensiveCache )] );
}
- return array( null, null, null );
}
/**
final protected static function isValidContainerName( $container ) {
// This accounts for Swift and S3 restrictions while leaving room
// for things like '.xxx' (hex shard chars) or '.seg' (segments).
+ // This disallows directory separators or traversal characters.
// Note that matching strings URL encode to the same string;
- // in Swift, the length resriction is *after* URL encoding.
+ // in Swift, the length restriction is *after* URL encoding.
return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
}
- /**
- * Validate and normalize a relative storage path.
- * Null is returned if the path involves directory traversal.
- * Traversal is insecure for FS backends and broken for others.
- *
- * @param $path string Storage path relative to a container
- * @return string|null
- */
- final protected static function normalizeContainerPath( $path ) {
- // Normalize directory separators
- $path = strtr( $path, '\\', '/' );
- // Collapse consecutive directory separators
- $path = preg_replace( '![/]{2,}!', '/', $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;
- }
-
/**
* Splits a storage path into an internal container name,
* an internal relative file name, and a container shard suffix.
* Like resolveStoragePath() except null values are returned if
* the container is sharded and the shard could not be determined.
*
- * @see FileBackend::resolveStoragePath()
+ * @see FileBackendStore::resolveStoragePath()
*
* @param $storagePath string
* @return Array (container, path) or (null, null) if invalid
* @return string|null Returns null if shard could not be determined
*/
final protected function getContainerShard( $container, $relPath ) {
- $hashLevels = $this->getContainerHashLevels( $container );
- if ( $hashLevels === 1 ) { // 16 shards per container
- $hashDirRegex = '(?P<shard>[0-9a-f])';
- } elseif ( $hashLevels === 2 ) { // 256 shards per container
- $hashDirRegex = '[0-9a-f]/(?P<shard>[0-9a-f]{2})';
- } else {
- return ''; // no sharding
- }
- // Allow certain directories to be above the hash dirs so as
- // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
- // They must be 2+ chars to avoid any hash directory ambiguity.
- if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
- return '.' . str_pad( $m['shard'], $hashLevels, '0', STR_PAD_LEFT );
+ 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 null; // failed to match
+ return ''; // no sharding
}
/**
- * Get the number of hash levels for a container.
+ * Get the sharding config for a container.
* If greater than 0, then all file storage paths within
* the container are required to be hashed accordingly.
*
* @param $container string
- * @return integer
+ * @return Array (integer levels, integer base, repeat flag) or (0, 0, false)
*/
final protected function getContainerHashLevels( $container ) {
if ( isset( $this->shardViaHashLevels[$container] ) ) {
- $hashLevels = (int)$this->shardViaHashLevels[$container];
- if ( $hashLevels >= 0 && $hashLevels <= 2 ) {
- return $hashLevels;
+ $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 0; // no sharding
+ return array( 0, 0, false ); // no sharding
}
/**
*/
final protected function getContainerSuffixes( $container ) {
$shards = array();
- $digits = $this->getContainerHashLevels( $container );
+ list( $digits, $base ) = $this->getContainerHashLevels( $container );
if ( $digits > 0 ) {
- $numShards = 1 << ( $digits * 4 );
+ $numShards = pow( $base, $digits );
for ( $index = 0; $index < $numShards; $index++ ) {
- $shards[] = '.' . str_pad( dechex( $index ), $digits, '0', STR_PAD_LEFT );
+ $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
}
}
return $shards;
protected function resolveContainerPath( $container, $relStoragePath ) {
return $relStoragePath;
}
-
- /**
- * Get the final extension from a storage or FS path
- *
- * @param $path string
- * @return string
- */
- final public static function extensionFromPath( $path ) {
- $i = strrpos( $path, '.' );
- return strtolower( $i ? substr( $path, $i + 1 ) : '' );
- }
}
/**
- * FileBackend helper function to handle file listings that span container shards.
- * Do not use this class from places outside of FileBackend.
+ * FileBackendStore helper function to handle file listings that span container shards.
+ * Do not use this class from places outside of FileBackendStore.
*
* @ingroup FileBackend
*/
-class FileBackendShardListIterator implements Iterator {
- /* @var FileBackend */
+class FileBackendStoreShardListIterator implements Iterator {
+ /* @var FileBackendStore */
protected $backend;
/* @var Array */
protected $params;
protected $pos = 0; // integer
/**
- * @param $backend FileBackend
+ * @param $backend FileBackendStore
* @param $container string Full storage container name
* @param $dir string Storage directory relative to container
* @param $suffixes Array List of container shard suffixes
* @param $params Array
*/
public function __construct(
- FileBackend $backend, $container, $dir, array $suffixes, array $params
+ FileBackendStore $backend, $container, $dir, array $suffixes, array $params
) {
$this->backend = $backend;
$this->container = $container;