# Copyright 2011-present MongoDB, 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. """Framework-agnostic core of Motor, an asynchronous driver for MongoDB.""" import functools import sys import warnings import pymongo import pymongo.auth import pymongo.common import pymongo.database import pymongo.errors import pymongo.mongo_client import pymongo.mongo_replica_set_client import pymongo.monotonic from pymongo.change_stream import ChangeStream from pymongo.client_session import ClientSession from pymongo.collection import Collection from pymongo.command_cursor import CommandCursor, RawBatchCommandCursor from pymongo.cursor import Cursor, RawBatchCursor, _QUERY_OPTIONS from pymongo.database import Database from pymongo.driver_info import DriverInfo from . import version as motor_version from .metaprogramming import (AsyncCommand, AsyncRead, AsyncWrite, coroutine_annotation, create_class_with_framework, DelegateMethod, MotorCursorChainingMethod, ReadOnlyProperty, unwrap_args_session, unwrap_kwargs_session) from .motor_common import callback_type_error from motor.docstrings import * HAS_SSL = True try: import ssl except ImportError: ssl = None HAS_SSL = False # From the Convenient API for Transactions spec, with_transaction must # halt retries after 120 seconds. # This limit is non-configurable and was chosen to be twice the 60 second # default value of MongoDB's `transactionLifetimeLimitSeconds` parameter. _WITH_TRANSACTION_RETRY_TIME_LIMIT = 120 def _within_time_limit(start_time): """Are we within the with_transaction retry limit?""" return (pymongo.monotonic.time() - start_time < _WITH_TRANSACTION_RETRY_TIME_LIMIT) def _max_time_expired_error(exc): """Return true if exc is a MaxTimeMSExpired error.""" return isinstance(exc, pymongo.errors.OperationFailure) and exc.code == 50 class AgnosticBase(object): def __eq__(self, other): if (isinstance(other, self.__class__) and hasattr(self, 'delegate') and hasattr(other, 'delegate')): return self.delegate == other.delegate return NotImplemented def __init__(self, delegate): self.delegate = delegate def __repr__(self): return '%s(%r)' % (self.__class__.__name__, self.delegate) class AgnosticBaseProperties(AgnosticBase): codec_options = ReadOnlyProperty() read_preference = ReadOnlyProperty() read_concern = ReadOnlyProperty() write_concern = ReadOnlyProperty() class AgnosticClient(AgnosticBaseProperties): __motor_class_name__ = 'MotorClient' __delegate_class__ = pymongo.mongo_client.MongoClient address = ReadOnlyProperty() arbiters = ReadOnlyProperty() close = DelegateMethod() drop_database = AsyncCommand().unwrap('MotorDatabase') event_listeners = ReadOnlyProperty() fsync = AsyncCommand(doc=fsync_doc) get_database = DelegateMethod(doc=get_database_doc).wrap(Database) get_default_database = DelegateMethod(doc=get_default_database_doc).wrap(Database) HOST = ReadOnlyProperty() is_mongos = ReadOnlyProperty() is_primary = ReadOnlyProperty() list_databases = AsyncRead().wrap(CommandCursor) list_database_names = AsyncRead() local_threshold_ms = ReadOnlyProperty() max_bson_size = ReadOnlyProperty() max_idle_time_ms = ReadOnlyProperty() max_message_size = ReadOnlyProperty() max_pool_size = ReadOnlyProperty() max_write_batch_size = ReadOnlyProperty() min_pool_size = ReadOnlyProperty() nodes = ReadOnlyProperty() PORT = ReadOnlyProperty() primary = ReadOnlyProperty() read_concern = ReadOnlyProperty() retry_reads = ReadOnlyProperty() retry_writes = ReadOnlyProperty() secondaries = ReadOnlyProperty() server_info = AsyncRead() server_selection_timeout = ReadOnlyProperty() start_session = AsyncCommand(doc=start_session_doc).wrap(ClientSession) unlock = AsyncCommand(doc=unlock_doc) def __init__(self, *args, **kwargs): """Create a new connection to a single MongoDB instance at *host:port*. Takes the same constructor arguments as :class:`~pymongo.mongo_client.MongoClient`, as well as: :Parameters: - `io_loop` (optional): Special event loop instance to use instead of default """ if 'io_loop' in kwargs: io_loop = kwargs.pop('io_loop') self._framework.check_event_loop(io_loop) else: io_loop = self._framework.get_event_loop() kwargs.setdefault('connect', False) kwargs.setdefault( 'driver', DriverInfo('Motor', motor_version, self._framework.platform_info())) delegate = self.__delegate_class__(*args, **kwargs) super().__init__(delegate) self.io_loop = io_loop def get_io_loop(self): return self.io_loop def watch(self, pipeline=None, full_document=None, resume_after=None, max_await_time_ms=None, batch_size=None, collation=None, start_at_operation_time=None, session=None, start_after=None): """Watch changes on this cluster. Returns a :class:`~MotorChangeStream` cursor which iterates over changes on all databases in this cluster. Introduced in MongoDB 4.0. See the documentation for :meth:`MotorCollection.watch` for more details and examples. :Parameters: - `pipeline` (optional): A list of aggregation pipeline stages to append to an initial ``$changeStream`` stage. Not all pipeline stages are valid after a ``$changeStream`` stage, see the MongoDB documentation on change streams for the supported stages. - `full_document` (optional): The fullDocument option to pass to the ``$changeStream`` stage. Allowed values: 'updateLookup'. When set to 'updateLookup', the change notification for partial updates will include both a delta describing the changes to the document, as well as a copy of the entire document that was changed from some time after the change occurred. - `resume_after` (optional): A resume token. If provided, the change stream will start returning changes that occur directly after the operation specified in the resume token. A resume token is the _id value of a change document. - `max_await_time_ms` (optional): The maximum time in milliseconds for the server to wait for changes before responding to a getMore operation. - `batch_size` (optional): The maximum number of documents to return per batch. - `collation` (optional): The :class:`~pymongo.collation.Collation` to use for the aggregation. - `start_at_operation_time` (optional): If provided, the resulting change stream will only return changes that occurred at or after the specified :class:`~bson.timestamp.Timestamp`. Requires MongoDB >= 4.0. - `session` (optional): a :class:`~pymongo.client_session.ClientSession`. - `start_after` (optional): The same as `resume_after` except that `start_after` can resume notifications after an invalidate event. This option and `resume_after` are mutually exclusive. :Returns: A :class:`~MotorChangeStream`. .. versionchanged:: 2.1 Added the ``start_after`` parameter. .. versionadded:: 2.0 .. mongodoc:: changeStreams """ cursor_class = create_class_with_framework( AgnosticChangeStream, self._framework, self.__module__) # Latent cursor that will send initial command on first "async for". return cursor_class(self, pipeline, full_document, resume_after, max_await_time_ms, batch_size, collation, start_at_operation_time, session, start_after) def __getattr__(self, name): if name.startswith('_'): raise AttributeError( "%s has no attribute %r. To access the %s" " database, use client['%s']." % ( self.__class__.__name__, name, name, name)) return self[name] def __getitem__(self, name): db_class = create_class_with_framework( AgnosticDatabase, self._framework, self.__module__) return db_class(self, name) def wrap(self, obj): if obj.__class__ == Database: db_class = create_class_with_framework( AgnosticDatabase, self._framework, self.__module__) return db_class(self, obj.name, _delegate=obj) elif obj.__class__ == CommandCursor: command_cursor_class = create_class_with_framework( AgnosticCommandCursor, self._framework, self.__module__) return command_cursor_class(obj, self) elif obj.__class__ == ClientSession: session_class = create_class_with_framework( AgnosticClientSession, self._framework, self.__module__) return session_class(obj, self) class _MotorTransactionContext(object): """Internal transaction context manager for start_transaction.""" def __init__(self, session): self._session = session async def __aenter__(self): return self async def __aexit__(self, exc_type, exc_val, exc_tb): if self._session.in_transaction: if exc_val is None: await self._session.commit_transaction() else: await self._session.abort_transaction() class AgnosticClientSession(AgnosticBase): """A session for ordering sequential operations. Do not create an instance of :class:`MotorClientSession` directly; use :meth:`MotorClient.start_session`: .. code-block:: python3 collection = client.db.collection async with await client.start_session() as s: async with s.start_transaction(): await collection.delete_one({'x': 1}, session=s) await collection.insert_one({'x': 2}, session=s) .. versionadded:: 2.0 """ __motor_class_name__ = 'MotorClientSession' __delegate_class__ = ClientSession commit_transaction = AsyncCommand() abort_transaction = AsyncCommand() end_session = AsyncCommand() cluster_time = ReadOnlyProperty() has_ended = ReadOnlyProperty() in_transaction = ReadOnlyProperty() options = ReadOnlyProperty() operation_time = ReadOnlyProperty() session_id = ReadOnlyProperty() advance_cluster_time = DelegateMethod() advance_operation_time = DelegateMethod() def __init__(self, delegate, motor_client): AgnosticBase.__init__(self, delegate=delegate) self._client = motor_client def get_io_loop(self): return self._client.get_io_loop() async def with_transaction(self, coro, read_concern=None, write_concern=None, read_preference=None, max_commit_time_ms=None): """Executes an awaitable in a transaction. This method starts a transaction on this session, awaits ``coro`` once, and then commits the transaction. For example:: async def coro(session): orders = session.client.db.orders inventory = session.client.db.inventory inserted_id = await orders.insert_one( {"sku": "abc123", "qty": 100}, session=session) await inventory.update_one( {"sku": "abc123", "qty": {"$gte": 100}}, {"$inc": {"qty": -100}}, session=session) return inserted_id async with await client.start_session() as session: inserted_id = await session.with_transaction(coro) To pass arbitrary arguments to the ``coro``, wrap it with a ``lambda`` like this:: async def coro(session, custom_arg, custom_kwarg=None): # Transaction operations... async with await client.start_session() as session: await session.with_transaction( lambda s: coro(s, "custom_arg", custom_kwarg=1)) In the event of an exception, ``with_transaction`` may retry the commit or the entire transaction, therefore ``coro`` may be awaited multiple times by a single call to ``with_transaction``. Developers should be mindful of this possiblity when writing a ``coro`` that modifies application state or has any other side-effects. Note that even when the ``coro`` is invoked multiple times, ``with_transaction`` ensures that the transaction will be committed at-most-once on the server. The ``coro`` should not attempt to start new transactions, but should simply run operations meant to be contained within a transaction. The ``coro`` should also not commit the transaction; this is handled automatically by ``with_transaction``. If the ``coro`` does commit or abort the transaction without error, however, ``with_transaction`` will return without taking further action. When ``coro`` raises an exception, ``with_transaction`` automatically aborts the current transaction. When ``coro`` or :meth:`~ClientSession.commit_transaction` raises an exception that includes the ``"TransientTransactionError"`` error label, ``with_transaction`` starts a new transaction and re-executes the ``coro``. When :meth:`~ClientSession.commit_transaction` raises an exception with the ``"UnknownTransactionCommitResult"`` error label, ``with_transaction`` retries the commit until the result of the transaction is known. This method will cease retrying after 120 seconds has elapsed. This timeout is not configurable and any exception raised by the ``coro`` or by :meth:`ClientSession.commit_transaction` after the timeout is reached will be re-raised. Applications that desire a different timeout duration should not use this method. :Parameters: - `coro`: The coroutine to run inside a transaction. The coroutine must accept a single argument, this session. Note, under certain error conditions the coroutine may be run multiple times. - `read_concern` (optional): The :class:`~pymongo.read_concern.ReadConcern` to use for this transaction. - `write_concern` (optional): The :class:`~pymongo.write_concern.WriteConcern` to use for this transaction. - `read_preference` (optional): The read preference to use for this transaction. If ``None`` (the default) the :attr:`read_preference` of this :class:`Database` is used. See :mod:`~pymongo.read_preferences` for options. :Returns: The return value of the ``coro``. .. versionadded:: 2.1 """ start_time = pymongo.monotonic.time() while True: async with self.start_transaction( read_concern, write_concern, read_preference, max_commit_time_ms): try: ret = await coro(self) except Exception as exc: if self.in_transaction: await self.abort_transaction() if (isinstance(exc, pymongo.errors.PyMongoError) and exc.has_error_label("TransientTransactionError") and _within_time_limit(start_time)): # Retry the entire transaction. continue raise if not self.in_transaction: # Assume callback intentionally ended the transaction. return ret while True: try: await self.commit_transaction() except pymongo.errors.PyMongoError as exc: if (exc.has_error_label("UnknownTransactionCommitResult") and _within_time_limit(start_time) and not _max_time_expired_error(exc)): # Retry the commit. continue if (exc.has_error_label("TransientTransactionError") and _within_time_limit(start_time)): # Retry the entire transaction. break raise # Commit succeeded. return ret def start_transaction(self, read_concern=None, write_concern=None, read_preference=None, max_commit_time_ms=None): """Start a multi-statement transaction. Takes the same arguments as :class:`~pymongo.client_session.TransactionOptions`. Best used in a context manager block: .. code-block:: python3 # Use "await" for start_session, but not for start_transaction. async with await client.start_session() as s: async with s.start_transaction(): await collection.delete_one({'x': 1}, session=s) await collection.insert_one({'x': 2}, session=s) """ self.delegate.start_transaction(read_concern=read_concern, write_concern=write_concern, read_preference=read_preference, max_commit_time_ms=max_commit_time_ms) return _MotorTransactionContext(self) @property def client(self): """The :class:`~MotorClient` this session was created from. """ return self._client async def __aenter__(self): return self async def __aexit__(self, exc_type, exc_val, exc_tb): self.delegate.__exit__(exc_type, exc_val, exc_tb) def __enter__(self): raise AttributeError("Use Motor sessions like 'async with await" " client.start_session()', not 'with'") def __exit__(self, exc_type, exc_val, exc_tb): pass class AgnosticDatabase(AgnosticBaseProperties): __motor_class_name__ = 'MotorDatabase' __delegate_class__ = Database command = AsyncCommand(doc=cmd_doc) create_collection = AsyncCommand().wrap(Collection) current_op = AsyncRead(doc=current_op_doc) dereference = AsyncRead() drop_collection = AsyncCommand().unwrap('MotorCollection') get_collection = DelegateMethod().wrap(Collection) list_collection_names = AsyncRead(doc=list_collection_names_doc) list_collections = AsyncRead() name = ReadOnlyProperty() profiling_info = AsyncRead() profiling_level = AsyncRead() set_profiling_level = AsyncCommand() validate_collection = AsyncRead().unwrap('MotorCollection') with_options = DelegateMethod().wrap(Database) incoming_manipulators = ReadOnlyProperty() incoming_copying_manipulators = ReadOnlyProperty() outgoing_manipulators = ReadOnlyProperty() outgoing_copying_manipulators = ReadOnlyProperty() _async_aggregate = AsyncRead(attr_name='aggregate') def __init__(self, client, name, **kwargs): self._client = client delegate = kwargs.get('_delegate') or Database( client.delegate, name, **kwargs) super().__init__(delegate) def aggregate(self, pipeline, **kwargs): """Execute an aggregation pipeline on this database. Introduced in MongoDB 3.6. The aggregation can be run on a secondary if the client is connected to a replica set and its ``read_preference`` is not :attr:`PRIMARY`. The :meth:`aggregate` method obeys the :attr:`read_preference` of this :class:`MotorDatabase`, except when ``$out`` or ``$merge`` are used, in which case :attr:`PRIMARY` is used. All optional `aggregate command`_ parameters should be passed as keyword arguments to this method. Valid options include, but are not limited to: - `allowDiskUse` (bool): Enables writing to temporary files. When set to True, aggregation stages can write data to the _tmp subdirectory of the --dbpath directory. The default is False. - `maxTimeMS` (int): The maximum amount of time to allow the operation to run in milliseconds. - `batchSize` (int): The maximum number of documents to return per batch. Ignored if the connected mongod or mongos does not support returning aggregate results using a cursor. - `collation` (optional): An instance of :class:`~pymongo.collation.Collation`. Returns a :class:`MotorCommandCursor` that can be iterated like a cursor from :meth:`find`:: async def f(): # Lists all operations currently running on the server. pipeline = [{"$currentOp": {}}] async for operation in client.admin.aggregate(pipeline): print(operation) .. note:: This method does not support the 'explain' option. Please use :meth:`MotorDatabase.command` instead. .. note:: The :attr:`MotorDatabase.write_concern` of this database is automatically applied to this operation. .. versionadded:: 2.1 .. _aggregate command: https://docs.mongodb.com/manual/reference/command/aggregate """ cursor_class = create_class_with_framework( AgnosticLatentCommandCursor, self._framework, self.__module__) # Latent cursor that will send initial command on first "async for". return cursor_class(self["$cmd.aggregate"], self._async_aggregate, pipeline, **unwrap_kwargs_session(kwargs)) def watch(self, pipeline=None, full_document=None, resume_after=None, max_await_time_ms=None, batch_size=None, collation=None, start_at_operation_time=None, session=None, start_after=None): """Watch changes on this database. Returns a :class:`~MotorChangeStream` cursor which iterates over changes on this database. Introduced in MongoDB 4.0. See the documentation for :meth:`MotorCollection.watch` for more details and examples. :Parameters: - `pipeline` (optional): A list of aggregation pipeline stages to append to an initial ``$changeStream`` stage. Not all pipeline stages are valid after a ``$changeStream`` stage, see the MongoDB documentation on change streams for the supported stages. - `full_document` (optional): The fullDocument option to pass to the ``$changeStream`` stage. Allowed values: 'updateLookup'. When set to 'updateLookup', the change notification for partial updates will include both a delta describing the changes to the document, as well as a copy of the entire document that was changed from some time after the change occurred. - `resume_after` (optional): A resume token. If provided, the change stream will start returning changes that occur directly after the operation specified in the resume token. A resume token is the _id value of a change document. - `max_await_time_ms` (optional): The maximum time in milliseconds for the server to wait for changes before responding to a getMore operation. - `batch_size` (optional): The maximum number of documents to return per batch. - `collation` (optional): The :class:`~pymongo.collation.Collation` to use for the aggregation. - `start_at_operation_time` (optional): If provided, the resulting change stream will only return changes that occurred at or after the specified :class:`~bson.timestamp.Timestamp`. Requires MongoDB >= 4.0. - `session` (optional): a :class:`~pymongo.client_session.ClientSession`. - `start_after` (optional): The same as `resume_after` except that `start_after` can resume notifications after an invalidate event. This option and `resume_after` are mutually exclusive. :Returns: A :class:`~MotorChangeStream`. .. versionchanged:: 2.1 Added the ``start_after`` parameter. .. versionadded:: 2.0 .. mongodoc:: changeStreams """ cursor_class = create_class_with_framework( AgnosticChangeStream, self._framework, self.__module__) # Latent cursor that will send initial command on first "async for". return cursor_class(self, pipeline, full_document, resume_after, max_await_time_ms, batch_size, collation, start_at_operation_time, session, start_after) @property def client(self): """This MotorDatabase's :class:`MotorClient`.""" return self._client def __getattr__(self, name): if name.startswith('_'): raise AttributeError( "%s has no attribute %r. To access the %s" " collection, use database['%s']." % ( self.__class__.__name__, name, name, name)) return self[name] def __getitem__(self, name): collection_class = create_class_with_framework( AgnosticCollection, self._framework, self.__module__) return collection_class(self, name) def __call__(self, *args, **kwargs): database_name = self.delegate.name client_class_name = self._client.__class__.__name__ if database_name == 'open_sync': raise TypeError( "%s.open_sync() is unnecessary Motor 0.2, " "see changelog for details." % client_class_name) raise TypeError( "MotorDatabase object is not callable. If you meant to " "call the '%s' method on a %s object it is " "failing because no such method exists." % ( database_name, client_class_name)) def wrap(self, obj): if obj.__class__ is Collection: # Replace pymongo.collection.Collection with MotorCollection. klass = create_class_with_framework( AgnosticCollection, self._framework, self.__module__) return klass(self, obj.name, _delegate=obj) elif obj.__class__ is Database: return self.__class__(self._client, obj.name, _delegate=obj) else: return obj def get_io_loop(self): return self._client.get_io_loop() class AgnosticCollection(AgnosticBaseProperties): __motor_class_name__ = 'MotorCollection' __delegate_class__ = Collection bulk_write = AsyncCommand(doc=bulk_write_doc) count_documents = AsyncRead() create_index = AsyncCommand() create_indexes = AsyncCommand(doc=create_indexes_doc) delete_many = AsyncCommand(doc=delete_many_doc) delete_one = AsyncCommand(doc=delete_one_doc) distinct = AsyncRead() drop = AsyncCommand(doc=drop_doc) drop_index = AsyncCommand() drop_indexes = AsyncCommand() estimated_document_count = AsyncCommand() find_one = AsyncRead(doc=find_one_doc) find_one_and_delete = AsyncCommand(doc=find_one_and_delete_doc) find_one_and_replace = AsyncCommand(doc=find_one_and_replace_doc) find_one_and_update = AsyncCommand(doc=find_one_and_update_doc) full_name = ReadOnlyProperty() index_information = AsyncRead(doc=index_information_doc) inline_map_reduce = AsyncRead() insert_many = AsyncWrite(doc=insert_many_doc) insert_one = AsyncCommand(doc=insert_one_doc) map_reduce = AsyncCommand(doc=mr_doc).wrap(Collection) name = ReadOnlyProperty() options = AsyncRead() reindex = AsyncCommand(doc=reindex_doc) rename = AsyncCommand() replace_one = AsyncCommand(doc=replace_one_doc) update_many = AsyncCommand(doc=update_many_doc) update_one = AsyncCommand(doc=update_one_doc) with_options = DelegateMethod().wrap(Collection) _async_aggregate = AsyncRead(attr_name='aggregate') _async_aggregate_raw_batches = AsyncRead(attr_name='aggregate_raw_batches') _async_list_indexes = AsyncRead(attr_name='list_indexes') def __init__(self, database, name, codec_options=None, read_preference=None, write_concern=None, read_concern=None, _delegate=None): db_class = create_class_with_framework( AgnosticDatabase, self._framework, self.__module__) if not isinstance(database, db_class): raise TypeError("First argument to MotorCollection must be " "MotorDatabase, not %r" % database) delegate = _delegate or Collection( database.delegate, name, codec_options=codec_options, read_preference=read_preference, write_concern=write_concern, read_concern=read_concern) super().__init__(delegate) self.database = database def __getattr__(self, name): # Dotted collection name, like "foo.bar". if name.startswith('_'): full_name = "%s.%s" % (self.name, name) raise AttributeError( "%s has no attribute %r. To access the %s" " collection, use database['%s']." % ( self.__class__.__name__, name, full_name, full_name)) return self[name] def __getitem__(self, name): collection_class = create_class_with_framework( AgnosticCollection, self._framework, self.__module__) return collection_class(self.database, self.name + '.' + name, _delegate=self.delegate[name]) def __call__(self, *args, **kwargs): raise TypeError( "MotorCollection object is not callable. If you meant to " "call the '%s' method on a MotorCollection object it is " "failing because no such method exists." % self.delegate.name) def find(self, *args, **kwargs): """Create a :class:`MotorCursor`. Same parameters as for PyMongo's :meth:`~pymongo.collection.Collection.find`. Note that ``find`` does not require an ``await`` expression, because ``find`` merely creates a :class:`MotorCursor` without performing any operations on the server. ``MotorCursor`` methods such as :meth:`~MotorCursor.to_list` perform actual operations. """ cursor = self.delegate.find(*unwrap_args_session(args), **unwrap_kwargs_session(kwargs)) cursor_class = create_class_with_framework( AgnosticCursor, self._framework, self.__module__) return cursor_class(cursor, self) def find_raw_batches(self, *args, **kwargs): """Query the database and retrieve batches of raw BSON. Similar to the :meth:`find` method but returns each batch as bytes. This example demonstrates how to work with raw batches, but in practice raw batches should be passed to an external library that can decode BSON into another data type, rather than used with PyMongo's :mod:`bson` module. .. code-block:: python3 async def get_raw(): cursor = db.test.find_raw_batches() async for batch in cursor: print(bson.decode_all(batch)) Note that ``find_raw_batches`` does not support sessions. .. versionadded:: 2.0 """ cursor = self.delegate.find_raw_batches(*unwrap_args_session(args), **unwrap_kwargs_session(kwargs)) cursor_class = create_class_with_framework( AgnosticRawBatchCursor, self._framework, self.__module__) return cursor_class(cursor, self) def aggregate(self, pipeline, **kwargs): """Execute an aggregation pipeline on this collection. The aggregation can be run on a secondary if the client is connected to a replica set and its ``read_preference`` is not :attr:`PRIMARY`. :Parameters: - `pipeline`: a single command or list of aggregation commands - `session` (optional): a :class:`~pymongo.client_session.ClientSession`, created with :meth:`~MotorClient.start_session`. - `**kwargs`: send arbitrary parameters to the aggregate command Returns a :class:`MotorCommandCursor` that can be iterated like a cursor from :meth:`find`:: async def f(): pipeline = [{'$project': {'name': {'$toUpper': '$name'}}}] async for doc in collection.aggregate(pipeline): print(doc) :class:`MotorCommandCursor` does not allow the ``explain`` option. To explain MongoDB's query plan for the aggregation, use :meth:`MotorDatabase.command`:: async def f(): plan = await db.command( 'aggregate', 'COLLECTION-NAME', pipeline=[{'$project': {'x': 1}}], explain=True) print(plan) .. versionchanged:: 2.1 This collection's read concern is now applied to pipelines containing the `$out` stage when connected to MongoDB >= 4.2. .. versionchanged:: 1.0 :meth:`aggregate` now **always** returns a cursor. .. versionchanged:: 0.5 :meth:`aggregate` now returns a cursor by default, and the cursor is returned immediately without an ``await``. See :ref:`aggregation changes in Motor 0.5 `. .. versionchanged:: 0.2 Added cursor support. .. _aggregate command: http://docs.mongodb.org/manual/applications/aggregation """ cursor_class = create_class_with_framework( AgnosticLatentCommandCursor, self._framework, self.__module__) # Latent cursor that will send initial command on first "async for". return cursor_class(self, self._async_aggregate, pipeline, **unwrap_kwargs_session(kwargs)) def aggregate_raw_batches(self, pipeline, **kwargs): """Perform an aggregation and retrieve batches of raw BSON. Similar to the :meth:`aggregate` method but returns each batch as bytes. This example demonstrates how to work with raw batches, but in practice raw batches should be passed to an external library that can decode BSON into another data type, rather than used with PyMongo's :mod:`bson` module. .. code-block:: python3 async def get_raw(): cursor = db.test.aggregate_raw_batches() async for batch in cursor: print(bson.decode_all(batch)) Note that ``aggregate_raw_batches`` does not support sessions. .. versionadded:: 2.0 """ cursor_class = create_class_with_framework( AgnosticLatentCommandCursor, self._framework, self.__module__) # Latent cursor that will send initial command on first "async for". return cursor_class(self, self._async_aggregate_raw_batches, pipeline, **unwrap_kwargs_session(kwargs)) def watch(self, pipeline=None, full_document=None, resume_after=None, max_await_time_ms=None, batch_size=None, collation=None, start_at_operation_time=None, session=None, start_after=None): """Watch changes on this collection. Performs an aggregation with an implicit initial ``$changeStream`` stage and returns a :class:`~MotorChangeStream` cursor which iterates over changes on this collection. Introduced in MongoDB 3.6. A change stream continues waiting indefinitely for matching change events. Code like the following allows a program to cancel the change stream and exit. .. code-block:: python3 change_stream = None async def watch_collection(): global change_stream # Using the change stream in an "async with" block # ensures it is canceled promptly if your code breaks # from the loop or throws an exception. async with db.collection.watch() as change_stream: async for change in change_stream: print(change) # Tornado from tornado.ioloop import IOLoop def main(): loop = IOLoop.current() # Start watching collection for changes. loop.add_callback(watch_collection) try: loop.start() except KeyboardInterrupt: pass finally: if change_stream is not None: change_stream.close() # asyncio from asyncio import get_event_loop def main(): loop = get_event_loop() task = loop.create_task(watch_collection) try: loop.run_forever() except KeyboardInterrupt: pass finally: if change_stream is not None: change_stream.close() # Prevent "Task was destroyed but it is pending!" loop.run_until_complete(task) The :class:`~MotorChangeStream` async iterable blocks until the next change document is returned or an error is raised. If the :meth:`~MotorChangeStream.next` method encounters a network error when retrieving a batch from the server, it will automatically attempt to recreate the cursor such that no change events are missed. Any error encountered during the resume attempt indicates there may be an outage and will be raised. .. code-block:: python3 try: pipeline = [{'$match': {'operationType': 'insert'}}] async with db.collection.watch(pipeline) as stream: async for change in stream: print(change) except pymongo.errors.PyMongoError: # The ChangeStream encountered an unrecoverable error or the # resume attempt failed to recreate the cursor. logging.error('...') For a precise description of the resume process see the `change streams specification`_. :Parameters: - `pipeline` (optional): A list of aggregation pipeline stages to append to an initial ``$changeStream`` stage. Not all pipeline stages are valid after a ``$changeStream`` stage, see the MongoDB documentation on change streams for the supported stages. - `full_document` (optional): The fullDocument option to pass to the ``$changeStream`` stage. Allowed values: 'updateLookup'. When set to 'updateLookup', the change notification for partial updates will include both a delta describing the changes to the document, as well as a copy of the entire document that was changed from some time after the change occurred. - `resume_after` (optional): A resume token. If provided, the change stream will start returning changes that occur directly after the operation specified in the resume token. A resume token is the _id value of a change document. - `max_await_time_ms` (optional): The maximum time in milliseconds for the server to wait for changes before responding to a getMore operation. - `batch_size` (optional): The maximum number of documents to return per batch. - `collation` (optional): The :class:`~pymongo.collation.Collation` to use for the aggregation. - `session` (optional): a :class:`~pymongo.client_session.ClientSession`. - `start_after` (optional): The same as `resume_after` except that `start_after` can resume notifications after an invalidate event. This option and `resume_after` are mutually exclusive. :Returns: A :class:`~MotorChangeStream`. See the :ref:`tornado_change_stream_example`. .. versionchanged:: 2.1 Added the ``start_after`` parameter. .. versionadded:: 1.2 .. mongodoc:: changeStreams .. _change streams specification: https://github.com/mongodb/specifications/blob/master/source/change-streams.rst """ cursor_class = create_class_with_framework( AgnosticChangeStream, self._framework, self.__module__) # Latent cursor that will send initial command on first "async for". return cursor_class(self, pipeline, full_document, resume_after, max_await_time_ms, batch_size, collation, start_at_operation_time, session, start_after) def list_indexes(self, session=None): """Get a cursor over the index documents for this collection. :: async def print_indexes(): async for index in db.test.list_indexes(): print(index) If the only index is the default index on ``_id``, this might print:: SON([('v', 1), ('key', SON([('_id', 1)])), ('name', '_id_')]) """ cursor_class = create_class_with_framework( AgnosticLatentCommandCursor, self._framework, self.__module__) # Latent cursor that will send initial command on first "async for". return cursor_class(self, self._async_list_indexes, session=session) def wrap(self, obj): if obj.__class__ is Collection: # Replace pymongo.collection.Collection with MotorCollection. return self.__class__(self.database, obj.name, _delegate=obj) elif obj.__class__ is Cursor: return AgnosticCursor(obj, self) elif obj.__class__ is CommandCursor: command_cursor_class = create_class_with_framework( AgnosticCommandCursor, self._framework, self.__module__) return command_cursor_class(obj, self) elif obj.__class__ is ChangeStream: change_stream_class = create_class_with_framework( AgnosticChangeStream, self._framework, self.__module__) return change_stream_class(obj, self) else: return obj def get_io_loop(self): return self.database.get_io_loop() class AgnosticBaseCursor(AgnosticBase): """Base class for AgnosticCursor and AgnosticCommandCursor""" _refresh = AsyncRead() address = ReadOnlyProperty() cursor_id = ReadOnlyProperty() alive = ReadOnlyProperty() session = ReadOnlyProperty() def __init__(self, cursor, collection): """Don't construct a cursor yourself, but acquire one from methods like :meth:`MotorCollection.find` or :meth:`MotorCollection.aggregate`. .. note:: There is no need to manually close cursors; they are closed by the server after being fully iterated with :meth:`to_list`, :meth:`each`, or `async for`, or automatically closed by the client when the :class:`MotorCursor` is cleaned up by the garbage collector. """ # 'cursor' is a PyMongo Cursor, CommandCursor, or a _LatentCursor. super().__init__(delegate=cursor) self.collection = collection self.started = False self.closed = False # python.org/dev/peps/pep-0492/#api-design-and-implementation-revisions def __aiter__(self): return self async def next(self): """Advance the cursor. .. versionadded:: 2.2 """ if self.alive and (self._buffer_size() or await self._get_more()): return next(self.delegate) raise StopAsyncIteration __anext__ = next def _get_more(self): """Initial query or getMore. Returns a Future.""" if not self.alive: raise pymongo.errors.InvalidOperation( "Can't call get_more() on a MotorCursor that has been" " exhausted or killed.") self.started = True return self._refresh() @property @coroutine_annotation def fetch_next(self): """**DEPRECATED** - A Future used with `gen.coroutine`_ to asynchronously retrieve the next document in the result set, fetching a batch of documents from the server if necessary. Resolves to ``False`` if there are no more documents, otherwise :meth:`next_object` is guaranteed to return a document: .. doctest:: fetch_next :hide: >>> _ = MongoClient().test.test_collection.delete_many({}) >>> collection = MotorClient().test.test_collection .. attention:: The :attr:`fetch_next` property is deprecated and will be removed in Motor 3.0. Use `async for` to iterate elegantly and efficiently over :class:`MotorCursor` objects instead.: .. doctest:: fetch_next >>> async def f(): ... await collection.drop() ... await collection.insert_many([{'_id': i} for i in range(5)]) ... async for doc in collection.find(): ... sys.stdout.write(str(doc['_id']) + ', ') ... print('done') ... >>> IOLoop.current().run_sync(f) 0, 1, 2, 3, 4, done While it appears that fetch_next retrieves each document from the server individually, the cursor actually fetches documents efficiently in `large batches`_. Example usage: .. doctest:: fetch_next >>> async def f(): ... await collection.drop() ... await collection.insert_many([{'_id': i} for i in range(5)]) ... cursor = collection.find().sort([('_id', 1)]) ... while (await cursor.fetch_next): ... doc = cursor.next_object() ... sys.stdout.write(str(doc['_id']) + ', ') ... print('done') ... >>> IOLoop.current().run_sync(f) 0, 1, 2, 3, 4, done .. versionchanged:: 2.2 Deprecated. .. _`large batches`: https://docs.mongodb.com/manual/tutorial/iterate-a-cursor/#cursor-batches .. _`gen.coroutine`: http://tornadoweb.org/en/stable/gen.html """ warnings.warn("The fetch_next property is deprecated and will be " "removed in Motor 3.0. Use `async for` to iterate " "over Cursor objects instead.", DeprecationWarning, stacklevel=2) if not self._buffer_size() and self.alive: # Return the Future, which resolves to number of docs fetched or 0. return self._get_more() elif self._buffer_size(): future = self._framework.get_future(self.get_io_loop()) future.set_result(True) return future else: # Dead future = self._framework.get_future(self.get_io_loop()) future.set_result(False) return future def next_object(self): """**DEPRECATED** - Get a document from the most recently fetched batch, or ``None``. See :attr:`fetch_next`. The :meth:`next_object` method is deprecated and will be removed in Motor 3.0. Use `async for` to elegantly iterate over :class:`MotorCursor` objects instead. .. versionchanged:: 2.2 Deprecated. """ warnings.warn("The next_object method is deprecated and will be " "removed in Motor 3.0. Use Use `async for` to iterate " "over Cursor objects instead.", DeprecationWarning, stacklevel=2) if not self._buffer_size(): return None return next(self.delegate) def each(self, callback): """Iterates over all the documents for this cursor. :meth:`each` returns immediately, and `callback` is executed asynchronously for each document. `callback` is passed ``(None, None)`` when iteration is complete. Cancel iteration early by returning ``False`` from the callback. (Only ``False`` cancels iteration: returning ``None`` or 0 does not.) .. testsetup:: each from tornado.ioloop import IOLoop MongoClient().test.test_collection.delete_many({}) MongoClient().test.test_collection.insert_many( [{'_id': i} for i in range(5)]) collection = MotorClient().test.test_collection .. doctest:: each >>> def each(result, error): ... if error: ... raise error ... elif result: ... sys.stdout.write(str(result['_id']) + ', ') ... else: ... # Iteration complete ... IOLoop.current().stop() ... print('done') ... >>> cursor = collection.find().sort([('_id', 1)]) >>> cursor.each(callback=each) >>> IOLoop.current().start() 0, 1, 2, 3, 4, done .. note:: Unlike other Motor methods, ``each`` requires a callback and does not return a Future, so it cannot be used in a coroutine. ``async for`` and :meth:`to_list` are much easier to use. :Parameters: - `callback`: function taking (document, error) """ if not callable(callback): raise callback_type_error self._each_got_more(callback, None) def _each_got_more(self, callback, future): if future: try: future.result() except Exception as error: callback(None, error) return while self._buffer_size() > 0: doc = next(self.delegate) # decrements self.buffer_size # Quit if callback returns exactly False (not None). Note we # don't close the cursor: user may want to resume iteration. if callback(doc, None) is False: return # The callback closed this cursor? if self.closed: return if self.alive and (self.cursor_id or not self.started): self._framework.add_future( self.get_io_loop(), self._get_more(), self._each_got_more, callback) else: # Complete self._framework.call_soon( self.get_io_loop(), functools.partial(callback, None, None)) @coroutine_annotation def to_list(self, length): """Get a list of documents. .. testsetup:: to_list MongoClient().test.test_collection.delete_many({}) MongoClient().test.test_collection.insert_many([{'_id': i} for i in range(4)]) from tornado import ioloop .. doctest:: to_list >>> from motor.motor_tornado import MotorClient >>> collection = MotorClient().test.test_collection >>> >>> async def f(): ... cursor = collection.find().sort([('_id', 1)]) ... docs = await cursor.to_list(length=2) ... while docs: ... print(docs) ... docs = await cursor.to_list(length=2) ... ... print('done') ... >>> ioloop.IOLoop.current().run_sync(f) [{'_id': 0}, {'_id': 1}] [{'_id': 2}, {'_id': 3}] done :Parameters: - `length`: maximum number of documents to return for this call, or None Returns a Future. .. versionchanged:: 2.0 No longer accepts a callback argument. .. versionchanged:: 0.2 `callback` must be passed as a keyword argument, like ``to_list(10, callback=callback)``, and the `length` parameter is no longer optional. """ if length is not None: if not isinstance(length, int): raise TypeError('length must be an int, not %r' % length) elif length < 0: raise ValueError('length must be non-negative') if self._query_flags() & _QUERY_OPTIONS['tailable_cursor']: raise pymongo.errors.InvalidOperation( "Can't call to_list on tailable cursor") future = self._framework.get_future(self.get_io_loop()) if not self.alive: future.set_result([]) else: the_list = [] self._framework.add_future( self.get_io_loop(), self._get_more(), self._to_list, length, the_list, future) return future def _to_list(self, length, the_list, future, get_more_result): # get_more_result is the result of self._get_more(). # to_list_future will be the result of the user's to_list() call. try: result = get_more_result.result() # Return early if the task was cancelled. if future.done(): return collection = self.collection fix_outgoing = collection.database.delegate._fix_outgoing if length is None: n = result else: n = min(length, result) for _ in range(n): the_list.append(fix_outgoing(self._data().popleft(), collection)) reached_length = (length is not None and len(the_list) >= length) if reached_length or not self.alive: future.set_result(the_list) else: self._framework.add_future( self.get_io_loop(), self._get_more(), self._to_list, length, the_list, future) except Exception as exc: if not future.done(): future.set_exception(exc) def get_io_loop(self): return self.collection.get_io_loop() async def close(self): """Explicitly kill this cursor on the server. Call like:: await cursor.close() """ if not self.closed: self.closed = True return self._close() def batch_size(self, batch_size): self.delegate.batch_size(batch_size) return self def _buffer_size(self): return len(self._data()) # Paper over some differences between PyMongo Cursor and CommandCursor. def _query_flags(self): raise NotImplementedError def _data(self): raise NotImplementedError def _clear_cursor_id(self): raise NotImplementedError def _close_exhaust_cursor(self): raise NotImplementedError def _killed(self): raise NotImplementedError async def _close(self): raise NotImplementedError() class AgnosticCursor(AgnosticBaseCursor): __motor_class_name__ = 'MotorCursor' __delegate_class__ = Cursor address = ReadOnlyProperty() collation = MotorCursorChainingMethod() distinct = AsyncRead() explain = AsyncRead() add_option = MotorCursorChainingMethod() remove_option = MotorCursorChainingMethod() limit = MotorCursorChainingMethod() skip = MotorCursorChainingMethod() max_scan = MotorCursorChainingMethod() sort = MotorCursorChainingMethod(doc=cursor_sort_doc) hint = MotorCursorChainingMethod() where = MotorCursorChainingMethod(doc=where_doc) max_await_time_ms = MotorCursorChainingMethod() max_time_ms = MotorCursorChainingMethod() min = MotorCursorChainingMethod() max = MotorCursorChainingMethod() comment = MotorCursorChainingMethod() allow_disk_use = MotorCursorChainingMethod() _Cursor__die = AsyncRead() def rewind(self): """Rewind this cursor to its unevaluated state.""" self.delegate.rewind() self.started = False return self def clone(self): """Get a clone of this cursor.""" return self.__class__(self.delegate.clone(), self.collection) def __copy__(self): return self.__class__(self.delegate.__copy__(), self.collection) def __deepcopy__(self, memo): return self.__class__(self.delegate.__deepcopy__(memo), self.collection) def _query_flags(self): return self.delegate._Cursor__query_flags def _data(self): return self.delegate._Cursor__data def _clear_cursor_id(self): self.delegate._Cursor__id = 0 def _close_exhaust_cursor(self): # If an exhaust cursor is dying without fully iterating its results, # it must close the socket. PyMongo's Cursor does this, but we've # disabled its cleanup so we must do it ourselves. if self.delegate._Cursor__exhaust: manager = self.delegate._Cursor__exhaust_mgr if manager.sock: manager.sock.close() manager.close() def _killed(self): return self.delegate._Cursor__killed def _close(self): return self._Cursor__die() class AgnosticRawBatchCursor(AgnosticCursor): __motor_class_name__ = 'MotorRawBatchCursor' __delegate_class__ = RawBatchCursor class AgnosticCommandCursor(AgnosticBaseCursor): __motor_class_name__ = 'MotorCommandCursor' __delegate_class__ = CommandCursor _CommandCursor__die = AsyncRead() def _query_flags(self): return 0 def _data(self): return self.delegate._CommandCursor__data def _clear_cursor_id(self): self.delegate._CommandCursor__id = 0 def _close_exhaust_cursor(self): # MongoDB doesn't have exhaust command cursors yet. pass def _killed(self): return self.delegate._CommandCursor__killed def _close(self): return self._CommandCursor__die() class AgnosticRawBatchCommandCursor(AgnosticCommandCursor): __motor_class_name__ = 'MotorRawBatchCommandCursor' __delegate_class__ = RawBatchCommandCursor class _LatentCursor(object): """Take the place of a PyMongo CommandCursor until aggregate() begins.""" alive = True _CommandCursor__data = [] _CommandCursor__id = None _CommandCursor__killed = False cursor_id = None def _CommandCursor__end_session(self, *args, **kwargs): pass def _CommandCursor__die(self, *args, **kwargs): pass def clone(self): return _LatentCursor() def rewind(self): pass class AgnosticLatentCommandCursor(AgnosticCommandCursor): __motor_class_name__ = 'MotorLatentCommandCursor' def __init__(self, collection, start, *args, **kwargs): # We're being constructed without await, like: # # cursor = collection.aggregate(pipeline) # # ... so we can't send the "aggregate" command to the server and get # a PyMongo CommandCursor back yet. Set self.delegate to a latent # cursor until the first await triggers _get_more(), which # will execute the callback "start", which gets a PyMongo CommandCursor. super().__init__(_LatentCursor(), collection) self.start = start self.args = args self.kwargs = kwargs def batch_size(self, batch_size): self.kwargs['batchSize'] = batch_size return self def _get_more(self): if not self.started: self.started = True original_future = self._framework.get_future(self.get_io_loop()) future = self.start( *self.args, **self.kwargs) self.start = self.args = self.kwargs = None self._framework.add_future( self.get_io_loop(), future, self._on_started, original_future) return original_future return super()._get_more() def _on_started(self, original_future, future): try: # "result" is a PyMongo command cursor from PyMongo's aggregate() or # aggregate_raw_batches(). Set its batch size from our latent # cursor's batch size. pymongo_cursor = future.result() self.delegate = pymongo_cursor except Exception as exc: if not original_future.done(): original_future.set_exception(exc) else: # Return early if the task was cancelled. if original_future.done(): return if self.delegate._CommandCursor__data or not self.delegate.alive: # _get_more is complete. original_future.set_result( len(self.delegate._CommandCursor__data)) else: # Send a getMore. future = super()._get_more() self._framework.chain_future(future, original_future) class AgnosticChangeStream(AgnosticBase): """A change stream cursor. Should not be called directly by application developers. See :meth:`~MotorCollection.watch` for example usage. .. versionadded: 1.2 .. mongodoc:: changeStreams """ __delegate_class__ = ChangeStream __motor_class_name__ = 'MotorChangeStream' _close = AsyncCommand(attr_name='close') alive = ReadOnlyProperty() resume_token = ReadOnlyProperty() def __init__(self, target, pipeline, full_document, resume_after, max_await_time_ms, batch_size, collation, start_at_operation_time, session, start_after): super().__init__(delegate=None) # The "target" object is a client, database, or collection. self._target = target self._kwargs = {'pipeline': pipeline, 'full_document': full_document, 'resume_after': resume_after, 'max_await_time_ms': max_await_time_ms, 'batch_size': batch_size, 'collation': collation, 'start_at_operation_time': start_at_operation_time, 'session': session, 'start_after': start_after} def _lazy_init(self): if not self.delegate: self.delegate = self._target.delegate.watch(**self._kwargs) def _next(self): # This method is run on a thread. try: self._lazy_init() return self.delegate.next() except StopIteration: raise StopAsyncIteration() def _try_next(self): # This method is run on a thread. self._lazy_init() return self.delegate.try_next() async def next(self): """Advance the cursor. This method blocks until the next change document is returned or an unrecoverable error is raised. This method is used when iterating over all changes in the cursor. For example:: async def watch_collection(): resume_token = None pipeline = [{'$match': {'operationType': 'insert'}}] try: async with db.collection.watch(pipeline) as stream: async for insert_change in stream: print(insert_change) resume_token = stream.resume_token except pymongo.errors.PyMongoError: # The ChangeStream encountered an unrecoverable error or the # resume attempt failed to recreate the cursor. if resume_token is None: # There is no usable resume token because there was a # failure during ChangeStream initialization. logging.error('...') else: # Use the interrupted ChangeStream's resume token to # create a new ChangeStream. The new stream will # continue from the last seen insert change without # missing any events. async with db.collection.watch( pipeline, resume_after=resume_token) as stream: async for insert_change in stream: print(insert_change) Raises :exc:`StopAsyncIteration` if this change stream is closed. In addition to using an "async for" loop as shown in the code example above, you can also iterate the change stream by calling ``await change_stream.next()`` repeatedly. """ loop = self.get_io_loop() return await self._framework.run_on_executor(loop, self._next) async def try_next(self): """Advance the cursor without blocking indefinitely. This method returns the next change document without waiting indefinitely for the next change. If no changes are available, it returns None. For example: .. code-block:: python3 while change_stream.alive: change = await change_stream.try_next() # Note that the ChangeStream's resume token may be updated # even when no changes are returned. print("Current resume token: %r" % (change_stream.resume_token,)) if change is not None: print("Change document: %r" % (change,)) continue # We end up here when there are no recent changes. # Sleep for a while before trying again to avoid flooding # the server with getMore requests when no changes are # available. await asyncio.sleep(10) If no change document is cached locally then this method runs a single getMore command. If the getMore yields any documents, the next document is returned, otherwise, if the getMore returns no documents (because there have been no changes) then ``None`` is returned. :Returns: The next change document or ``None`` when no document is available after running a single getMore or when the cursor is closed. .. versionadded:: 2.1 """ loop = self.get_io_loop() return await self._framework.run_on_executor(loop, self._try_next) async def close(self): """Close this change stream. Stops any "async for" loops using this change stream. """ if self.delegate: await self._close() def __aiter__(self): return self __anext__ = next async def __aenter__(self): if not self.delegate: loop = self.get_io_loop() await self._framework.run_on_executor(loop, self._lazy_init) return self async def __aexit__(self, exc_type, exc_val, exc_tb): if self.delegate: await self.close() def get_io_loop(self): return self._target.get_io_loop() def __enter__(self): raise RuntimeError('Use a change stream in "async with", not "with"') def __exit__(self, exc_type, exc_val, exc_tb): pass