#!/usr/bin/env python # # Copyright 2007 Google Inc. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ In-memory persistent stub for the Python datastore API. Gets, queries, and searches are implemented as in-memory scans over all entities. Stores entities across sessions as pickled proto bufs in a single file. On startup, all entities are read from the file and loaded into memory. On every Put(), the file is wiped and all entities are written from scratch. Clients can also manually Read() and Write() the file themselves. """ import collections import logging import os import struct import sys import tempfile import threading import weakref import cPickle as pickle from google.appengine.api import apiproxy_stub from google.appengine.api import datastore from google.appengine.api import datastore_types from google.appengine.datastore import datastore_pb from google.appengine.datastore import datastore_stub_util from google.appengine.runtime import apiproxy_errors from google.net.proto import ProtocolBuffer from google.appengine.datastore import entity_pb datastore_pb.Query.__hash__ = lambda self: hash(self.Encode()) def _FinalElement(key): """Return final element of a key's path.""" return key.path().element_list()[-1] class _StoredEntity(object): """Simple wrapper around an entity (and its metadata) stored by the stub. Public properties: record: the original EntityRecord that was stored by the stub. encoded_protobuf: Encoded binary representation of entity protobuf, including a special property that holds the pickled metadata object. """ def __init__(self, record): """Create a _StoredEntity object and store an entity. Args: record: the EntityRecord to store. """ self.record = record entity = datastore_stub_util._ToStorageEntity(record) self.encoded_protobuf = entity.Encode() class KindPseudoKind(object): """Pseudo-kind for schema queries. Provides a Query method to perform the actual query. Public properties: name: the pseudo-kind name """ name = '__kind__' def Query(self, query, filters, orders): """Perform a query on this pseudo-kind. Args: query: the original datastore_pb.Query. filters: the filters from query. orders: the orders from query. Returns: (results, remaining_filters, remaining_orders) results is a list of entity_pb.EntityProto remaining_filters and remaining_orders are the filters and orders that should be applied in memory """ kind_range = datastore_stub_util.ParseKindQuery(query, filters, orders) app_namespace_str = datastore_types.EncodeAppIdNamespace( query.app(), query.name_space()) kinds = [] for app_namespace, kind in self._stub._GetAllEntities(): if app_namespace != app_namespace_str: continue kind = kind.decode('utf-8') if not kind_range.Contains(kind): continue kinds.append(datastore.Entity(self.name, name=kind, _app=query.app(), namespace=query.name_space())._ToPb()) return (kinds, [], []) class PropertyPseudoKind(object): """Pseudo-kind for schema queries. Provides a Query method to perform the actual query. Public properties: name: the pseudo-kind name """ name = '__property__' def Query(self, query, filters, orders): """Perform a query on this pseudo-kind. Args: query: the original datastore_pb.Query. filters: the filters from query. orders: the orders from query. Returns: (results, remaining_filters, remaining_orders) results is a list of entity_pb.EntityProto remaining_filters and remaining_orders are the filters and orders that should be applied in memory """ property_range = datastore_stub_util.ParsePropertyQuery(query, filters, orders) keys_only = query.keys_only() app_namespace_str = datastore_types.EncodeAppIdNamespace( query.app(), query.name_space()) properties = [] if keys_only: usekey = '__property__keys' else: usekey = '__property__' entities = self._stub._GetAllEntities() for app_namespace, kind in entities: if app_namespace != app_namespace_str: continue app_kind = (app_namespace_str, kind) kind = kind.decode('utf-8') (start_cmp, end_cmp) = property_range.MapExtremes( lambda extreme, inclusive, is_end: cmp(kind, extreme[0])) if not((start_cmp is None or start_cmp >= 0) and (end_cmp is None or end_cmp <= 0)): continue kind_properties = self._stub._GetSchemaCache(app_kind, usekey) if not kind_properties: kind_properties = [] kind_key = datastore_types.Key.from_path(KindPseudoKind.name, kind, _app=query.app(), namespace=query.name_space()) props = collections.defaultdict(set) for entity in entities[app_kind].values(): for prop in entity.record.entity.property_list(): prop_name = prop.name() if (prop_name in datastore_stub_util.GetInvisibleSpecialPropertyNames()): continue value_pb = prop.value() props[prop_name].add(datastore_types.GetPropertyValueTag(value_pb)) for prop in sorted(props): property_e = datastore.Entity(self.name, name=prop, parent=kind_key, _app=query.app(), namespace=query.name_space()) if not keys_only and props[prop]: property_e['property_representation'] = [ datastore_stub_util._PROPERTY_TYPE_NAMES[tag] for tag in sorted(props[prop])] kind_properties.append(property_e._ToPb()) self._stub._SetSchemaCache(app_kind, usekey, kind_properties) def InQuery(property_e): return property_range.Contains( (kind, _FinalElement(property_e.key()).name())) properties += filter(InQuery, kind_properties) return (properties, [], []) class NamespacePseudoKind(object): """Pseudo-kind for namespace queries. Provides a Query method to perform the actual query. Public properties: name: the pseudo-kind name """ name = '__namespace__' def Query(self, query, filters, orders): """Perform a query on this pseudo-kind. Args: query: the original datastore_pb.Query. filters: the filters from query. orders: the orders from query. Returns: (results, remaining_filters, remaining_orders) results is a list of entity_pb.EntityProto remaining_filters and remaining_orders are the filters and orders that should be applied in memory """ namespace_range = datastore_stub_util.ParseNamespaceQuery(query, filters, orders) app_str = query.app() namespaces = set() for app_namespace, _ in self._stub._GetAllEntities(): (app_id, namespace) = datastore_types.DecodeAppIdNamespace(app_namespace) if app_id == app_str and namespace_range.Contains(namespace): namespaces.add(namespace) namespace_entities = [] for namespace in namespaces: if namespace: namespace_e = datastore.Entity(self.name, name=namespace, _app=query.app()) else: namespace_e = datastore.Entity(self.name, id=datastore_types._EMPTY_NAMESPACE_ID, _app=query.app()) namespace_entities.append(namespace_e._ToPb()) return (namespace_entities, [], []) class DatastoreFileStub(datastore_stub_util.BaseDatastore, apiproxy_stub.APIProxyStub, datastore_stub_util.DatastoreStub): """ Persistent stub for the Python datastore API. Stores all entities in memory, and persists them to a file as pickled protocol buffers. A DatastoreFileStub instance handles a single app's data and is backed by files on disk. """ def __init__(self, app_id, datastore_file, history_file=None, require_indexes=False, service_name='datastore_v3', trusted=False, consistency_policy=None, save_changes=True, root_path=None, use_atexit=True, auto_id_policy=datastore_stub_util.SEQUENTIAL): """Constructor. Initializes and loads the datastore from the backing files, if they exist. Args: app_id: string datastore_file: string, stores all entities across sessions. Use None not to use a file. history_file: DEPRECATED. No-op. require_indexes: bool, default False. If True, composite indexes must exist in index.yaml for queries that need them. service_name: Service name expected for all calls. trusted: bool, default False. If True, this stub allows an app to access the data of another app. consistency_policy: The consistency policy to use or None to use the default. Consistency policies can be found in datastore_stub_util.*ConsistencyPolicy save_changes: bool, default True. If this stub should modify datastore_file when entities are changed. root_path: string, the root path of the app. use_atexit: bool, indicates if the stub should save itself atexit. auto_id_policy: enum, datastore_stub_util.SEQUENTIAL or .SCATTERED """ self.__datastore_file = datastore_file self.__save_changes = save_changes self.__entities_by_kind = collections.defaultdict(dict) self.__entities_by_group = collections.defaultdict(dict) self.__entities_lock = threading.Lock() self.__schema_cache = {} self.__id_counters = {datastore_stub_util.SEQUENTIAL: 1L, datastore_stub_util.SCATTERED: 1L } self.__id_lock = threading.Lock() self.__file_lock = threading.Lock() datastore_stub_util.BaseDatastore.__init__( self, require_indexes, consistency_policy, use_atexit and self.__IsSaveable(), auto_id_policy) apiproxy_stub.APIProxyStub.__init__(self, service_name) datastore_stub_util.DatastoreStub.__init__(self, weakref.proxy(self), app_id, trusted, root_path) self._RegisterPseudoKind(KindPseudoKind()) self._RegisterPseudoKind(PropertyPseudoKind()) self._RegisterPseudoKind(NamespacePseudoKind()) self._RegisterPseudoKind(datastore_stub_util.EntityGroupPseudoKind()) self.Read() def Clear(self): """ Clears the datastore by deleting all currently stored entities and queries. """ self.__entities_lock.acquire() try: datastore_stub_util.BaseDatastore.Clear(self) datastore_stub_util.DatastoreStub.Clear(self) self.__entities_by_kind = collections.defaultdict(dict) self.__entities_by_group = collections.defaultdict(dict) self.__schema_cache = {} finally: self.__entities_lock.release() def _GetAllEntities(self): """Get all entities. Returns: Map from kind to _StoredEntity() list. Do not modify directly. """ return self.__entities_by_kind def _GetEntityLocation(self, key): """Get keys to self.__entities_by_* from the given key. Example usage: app_kind, eg_k, k = self._GetEntityLocation(key) self.__entities_by_kind[app_kind][k] self.__entities_by_entity_group[eg_k][k] Args: key: entity_pb.Reference Returns: Tuple (by_kind key, by_entity_group key, entity key) """ app_ns = datastore_types.EncodeAppIdNamespace(key.app(), key.name_space()) kind = _FinalElement(key).type() entity_group = datastore_stub_util._GetEntityGroup(key) eg_k = datastore_types.ReferenceToKeyValue(entity_group) k = datastore_types.ReferenceToKeyValue(key) return ((app_ns, kind), eg_k, k) def _StoreRecord(self, record, insert=False): """ Store the given entity record. Any needed locking should be managed by the caller. Args: record: The EntityRecord to store. insert: If we should check for existence. """ app_kind, eg_k, k = self._GetEntityLocation(record.entity.key()) assert not insert or k not in self.__entities_by_kind[app_kind] stored_entity = _StoredEntity(record) self.__entities_by_kind[app_kind][k] = stored_entity self.__entities_by_group[eg_k][k] = stored_entity if app_kind in self.__schema_cache: del self.__schema_cache[app_kind] READ_PB_EXCEPTIONS = (ProtocolBuffer.ProtocolBufferDecodeError, LookupError, TypeError, ValueError) READ_ERROR_MSG = ('Data in %s is corrupt or a different version. ' 'Try running with the --clear_datastore flag.\n%r') READ_PY250_MSG = ('Are you using FloatProperty and/or GeoPtProperty? ' 'Unfortunately loading float values from the datastore ' 'file does not work with Python 2.5.0. ' 'Please upgrade to a newer Python 2.5 release or use ' 'the --clear_datastore flag.\n') def Read(self): """ Reads the datastore and history files into memory. The in-memory query history is cleared, but the datastore is *not* cleared; the entities in the files are merged into the entities in memory. If you want them to overwrite the in-memory datastore, call Clear() before calling Read(). If the datastore file contains an entity with the same app name, kind, and key as an entity already in the datastore, the entity from the file overwrites the entity in the datastore. Also sets each ID counter to one greater than the highest ID allocated so far in that counter's ID space. """ if self.__datastore_file and self.__datastore_file != '/dev/null': entities = self.__ReadPickled(self.__datastore_file) if entities and isinstance(entities[-1], (int, long)): self._commit_timestamp = int(entities.pop()) for encoded_entity in entities: try: entity = entity_pb.EntityProto(encoded_entity) record = datastore_stub_util._FromStorageEntity(entity) except self.READ_PB_EXCEPTIONS, e: raise apiproxy_errors.ApplicationError( datastore_pb.Error.INTERNAL_ERROR, self.READ_ERROR_MSG % (self.__datastore_file, e)) except struct.error, e: if (sys.version_info[0:3] == (2, 5, 0) and e.message.startswith('unpack requires a string argument')): raise apiproxy_errors.ApplicationError( datastore_pb.Error.INTERNAL_ERROR, self.READ_PY250_MSG + self.READ_ERROR_MSG % (self.__datastore_file, e)) else: raise self._StoreRecord(record) last_path = _FinalElement(entity.key()) if last_path.id(): self._SetMaxId(last_path.id()) def Write(self): """Writes out the datastore and history files. Be careful! If the files already exist, this method overwrites them! """ super(DatastoreFileStub, self).Write() self.__WriteDatastore() def __IsSaveable(self): return (self.__datastore_file and self.__datastore_file != '/dev/null' and self.__save_changes) def __WriteDatastore(self): """ Writes out the datastore file. Be careful! If the file already exists, this method overwrites it! """ if self.__IsSaveable(): encoded = [] for kind_dict in self.__entities_by_kind.values(): encoded.extend(entity.encoded_protobuf for entity in kind_dict.values()) encoded.append(self._GetReadTimestamp()) self.__WritePickled(encoded, self.__datastore_file) def __ReadPickled(self, filename): """Reads a pickled object from the given file and returns it. """ self.__file_lock.acquire() try: try: if (filename and filename != '/dev/null' and os.path.isfile(filename) and os.stat(filename).st_size > 0): return pickle.load(open(filename, 'rb')) else: logging.warning('Could not read datastore data from %s', filename) except (AttributeError, LookupError, ImportError, NameError, TypeError, ValueError, struct.error, pickle.PickleError), e: raise apiproxy_errors.ApplicationError( datastore_pb.Error.INTERNAL_ERROR, 'Could not read data from %s. Try running with the ' '--clear_datastore flag. Cause:\n%r' % (filename, e)) finally: self.__file_lock.release() return [] def __WritePickled(self, obj, filename): """Pickles the object and writes it to the given file. """ if not filename or filename == '/dev/null' or not obj: return descriptor, tmp_filename = tempfile.mkstemp(dir=os.path.dirname(filename)) tmpfile = os.fdopen(descriptor, 'wb') pickler = pickle.Pickler(tmpfile, protocol=1) pickler.fast = True pickler.dump(obj) tmpfile.close() self.__file_lock.acquire() try: try: os.rename(tmp_filename, filename) except OSError: try: os.remove(filename) except: pass os.rename(tmp_filename, filename) finally: self.__file_lock.release() def MakeSyncCall(self, service, call, request, response, request_id=None): """ The main RPC entry point. service must be 'datastore_v3'.""" self.assertPbIsInitialized(request) super(DatastoreFileStub, self).MakeSyncCall(service, call, request, response, request_id) self.assertPbIsInitialized(response) def assertPbIsInitialized(self, pb): """Raises an exception if the given PB is not initialized and valid.""" explanation = [] assert pb.IsInitialized(explanation), explanation pb.Encode() def _GetSchemaCache(self, kind, usekey): if kind in self.__schema_cache and usekey in self.__schema_cache[kind]: return self.__schema_cache[kind][usekey] else: return None def _SetSchemaCache(self, kind, usekey, value): if kind not in self.__schema_cache: self.__schema_cache[kind] = {} self.__schema_cache[kind][usekey] = value def _Put(self, record, insert): record = datastore_stub_util.StoreRecord(record) self.__entities_lock.acquire() try: self._StoreRecord(record, insert) finally: self.__entities_lock.release() def _Get(self, key): app_kind, _, k = self._GetEntityLocation(key) try: stored_entity = self.__entities_by_kind[app_kind][k] return datastore_stub_util.LoadRecord(stored_entity.record) except KeyError: pass def _Delete(self, key): app_kind, eg_k, k = self._GetEntityLocation(key) self.__entities_lock.acquire() try: del self.__entities_by_kind[app_kind][k] del self.__entities_by_group[eg_k][k] if not self.__entities_by_kind[app_kind]: del self.__entities_by_kind[app_kind] if not self.__entities_by_group[eg_k]: del self.__entities_by_group[eg_k] del self.__schema_cache[app_kind] except KeyError: pass finally: self.__entities_lock.release() def _GetEntitiesInEntityGroup(self, entity_group): eg_k = datastore_types.ReferenceToKeyValue(entity_group) return dict((k, e.record) for (k, e) in self.__entities_by_group[eg_k].iteritems()) def _GetQueryCursor(self, query, filters, orders, index_list): app_id = query.app() namespace = query.name_space() pseudo_kind = None if query.has_kind() and query.kind() in self._pseudo_kinds: pseudo_kind = self._pseudo_kinds[query.kind()] self.__entities_lock.acquire() try: app_ns = datastore_types.EncodeAppIdNamespace(app_id, namespace) if pseudo_kind: (results, filters, orders) = pseudo_kind.Query(query, filters, orders) results = map(datastore_stub_util.EntityRecord, results) elif query.has_kind(): stored_entities = self.__entities_by_kind[app_ns, query.kind()].values() results = [stored_entity.record for stored_entity in stored_entities] else: results = [] for (cur_app_ns, _), entities in self.__entities_by_kind.iteritems(): if cur_app_ns == app_ns: results.extend(entity.record for entity in entities.itervalues()) except KeyError: results = [] finally: self.__entities_lock.release() return datastore_stub_util._ExecuteQuery(results, query, filters, orders, index_list) def _SetIdCounter(self, id_space, value): """Set the ID counter for id_space to value.""" self.__id_counters[id_space] = value def _IdCounter(self, id_space): """Return current value of ID counter for id_space.""" return self.__id_counters[id_space] def _SetMaxId(self, max_id): """Infer ID space and advance corresponding counter.""" count, id_space = datastore_stub_util.IdToCounter(max_id) if count >= self._IdCounter(id_space): self._SetIdCounter(id_space, count + 1) def _AllocateSequentialIds(self, reference, size=1, max_id=None): datastore_stub_util.Check(not (size and max_id), 'Both size and max cannot be set.') self.__id_lock.acquire() try: id_space = datastore_stub_util.SEQUENTIAL start = self._IdCounter(id_space) if size: datastore_stub_util.Check(size > 0, 'Size must be greater than 0.') self._SetIdCounter(id_space, start + size) elif max_id: datastore_stub_util.Check(max_id >=0, 'Max must be greater than or equal to 0.') self._SetIdCounter(id_space, max(start, max_id + 1)) end = self._IdCounter(id_space) - 1 finally: self.__id_lock.release() return (start, end) def _AllocateIds(self, keys): self.__id_lock.acquire() full_keys = [] try: for key in keys: last_element = _FinalElement(key) if last_element.id() or last_element.has_name(): for el in key.path().element_list(): if el.id(): self._SetMaxId(el.id()) else: id_space = datastore_stub_util.SCATTERED count = self._IdCounter(id_space) last_element.set_id(datastore_stub_util.ToScatteredId(count)) self._SetIdCounter(id_space, count + 1) full_keys.append(key) finally: self.__id_lock.release() return full_keys