import asyncio import engineio from . import async_manager from . import base_server from . import exceptions from . import packet # this set is used to keep references to background tasks to prevent them from # being garbage collected mid-execution. Solution taken from # https://docs.python.org/3/library/asyncio-task.html#asyncio.create_task task_reference_holder = set() class AsyncServer(base_server.BaseServer): """A Socket.IO server for asyncio. This class implements a fully compliant Socket.IO web server with support for websocket and long-polling transports, compatible with the asyncio framework. :param client_manager: The client manager instance that will manage the client list. When this is omitted, the client list is stored in an in-memory structure, so the use of multiple connected servers is not possible. :param logger: To enable logging set to ``True`` or pass a logger object to use. To disable logging set to ``False``. Note that fatal errors are logged even when ``logger`` is ``False``. :param json: An alternative json module to use for encoding and decoding packets. Custom json modules must have ``dumps`` and ``loads`` functions that are compatible with the standard library versions. :param async_handlers: If set to ``True``, event handlers for a client are executed in separate threads. To run handlers for a client synchronously, set to ``False``. The default is ``True``. :param always_connect: When set to ``False``, new connections are provisory until the connect handler returns something other than ``False``, at which point they are accepted. When set to ``True``, connections are immediately accepted, and then if the connect handler returns ``False`` a disconnect is issued. Set to ``True`` if you need to emit events from the connect handler and your client is confused when it receives events before the connection acceptance. In any other case use the default of ``False``. :param namespaces: a list of namespaces that are accepted, in addition to any namespaces for which handlers have been defined. The default is `['/']`, which always accepts connections to the default namespace. Set to `'*'` to accept all namespaces. :param kwargs: Connection parameters for the underlying Engine.IO server. The Engine.IO configuration supports the following settings: :param async_mode: The asynchronous model to use. See the Deployment section in the documentation for a description of the available options. Valid async modes are "aiohttp", "sanic", "tornado" and "asgi". If this argument is not given, "aiohttp" is tried first, followed by "sanic", "tornado", and finally "asgi". The first async mode that has all its dependencies installed is the one that is chosen. :param ping_interval: The interval in seconds at which the server pings the client. The default is 25 seconds. For advanced control, a two element tuple can be given, where the first number is the ping interval and the second is a grace period added by the server. :param ping_timeout: The time in seconds that the client waits for the server to respond before disconnecting. The default is 20 seconds. :param max_http_buffer_size: The maximum size that is accepted for incoming messages. The default is 1,000,000 bytes. In spite of its name, the value set in this argument is enforced for HTTP long-polling and WebSocket connections. :param allow_upgrades: Whether to allow transport upgrades or not. The default is ``True``. :param http_compression: Whether to compress packages when using the polling transport. The default is ``True``. :param compression_threshold: Only compress messages when their byte size is greater than this value. The default is 1024 bytes. :param cookie: If set to a string, it is the name of the HTTP cookie the server sends back to the client containing the client session id. If set to a dictionary, the ``'name'`` key contains the cookie name and other keys define cookie attributes, where the value of each attribute can be a string, a callable with no arguments, or a boolean. If set to ``None`` (the default), a cookie is not sent to the client. :param cors_allowed_origins: Origin or list of origins that are allowed to connect to this server. Only the same origin is allowed by default. Set this argument to ``'*'`` to allow all origins, or to ``[]`` to disable CORS handling. :param cors_credentials: Whether credentials (cookies, authentication) are allowed in requests to this server. The default is ``True``. :param monitor_clients: If set to ``True``, a background task will ensure inactive clients are closed. Set to ``False`` to disable the monitoring task (not recommended). The default is ``True``. :param engineio_logger: To enable Engine.IO logging set to ``True`` or pass a logger object to use. To disable logging set to ``False``. The default is ``False``. Note that fatal errors are logged even when ``engineio_logger`` is ``False``. """ def __init__(self, client_manager=None, logger=False, json=None, async_handlers=True, namespaces=None, **kwargs): if client_manager is None: client_manager = async_manager.AsyncManager() super().__init__(client_manager=client_manager, logger=logger, json=json, async_handlers=async_handlers, namespaces=namespaces, **kwargs) def is_asyncio_based(self): return True def attach(self, app, socketio_path='socket.io'): """Attach the Socket.IO server to an application.""" self.eio.attach(app, socketio_path) async def emit(self, event, data=None, to=None, room=None, skip_sid=None, namespace=None, callback=None, ignore_queue=False): """Emit a custom event to one or more connected clients. :param event: The event name. It can be any string. The event names ``'connect'``, ``'message'`` and ``'disconnect'`` are reserved and should not be used. :param data: The data to send to the client or clients. Data can be of type ``str``, ``bytes``, ``list`` or ``dict``. To send multiple arguments, use a tuple where each element is of one of the types indicated above. :param to: The recipient of the message. This can be set to the session ID of a client to address only that client, to any any custom room created by the application to address all the clients in that room, or to a list of custom room names. If this argument is omitted the event is broadcasted to all connected clients. :param room: Alias for the ``to`` parameter. :param skip_sid: The session ID of a client to skip when broadcasting to a room or to all clients. This can be used to prevent a message from being sent to the sender. :param namespace: The Socket.IO namespace for the event. If this argument is omitted the event is emitted to the default namespace. :param callback: If given, this function will be called to acknowledge the client has received the message. The arguments that will be passed to the function are those provided by the client. Callback functions can only be used when addressing an individual client. :param ignore_queue: Only used when a message queue is configured. If set to ``True``, the event is emitted to the clients directly, without going through the queue. This is more efficient, but only works when a single server process is used. It is recommended to always leave this parameter with its default value of ``False``. Note: this method is not designed to be used concurrently. If multiple tasks are emitting at the same time to the same client connection, then messages composed of multiple packets may end up being sent in an incorrect sequence. Use standard concurrency solutions (such as a Lock object) to prevent this situation. Note 2: this method is a coroutine. """ namespace = namespace or '/' room = to or room self.logger.info('emitting event "%s" to %s [%s]', event, room or 'all', namespace) await self.manager.emit(event, data, namespace, room=room, skip_sid=skip_sid, callback=callback, ignore_queue=ignore_queue) async def send(self, data, to=None, room=None, skip_sid=None, namespace=None, callback=None, ignore_queue=False): """Send a message to one or more connected clients. This function emits an event with the name ``'message'``. Use :func:`emit` to issue custom event names. :param data: The data to send to the client or clients. Data can be of type ``str``, ``bytes``, ``list`` or ``dict``. To send multiple arguments, use a tuple where each element is of one of the types indicated above. :param to: The recipient of the message. This can be set to the session ID of a client to address only that client, to any any custom room created by the application to address all the clients in that room, or to a list of custom room names. If this argument is omitted the event is broadcasted to all connected clients. :param room: Alias for the ``to`` parameter. :param skip_sid: The session ID of a client to skip when broadcasting to a room or to all clients. This can be used to prevent a message from being sent to the sender. :param namespace: The Socket.IO namespace for the event. If this argument is omitted the event is emitted to the default namespace. :param callback: If given, this function will be called to acknowledge the client has received the message. The arguments that will be passed to the function are those provided by the client. Callback functions can only be used when addressing an individual client. :param ignore_queue: Only used when a message queue is configured. If set to ``True``, the event is emitted to the clients directly, without going through the queue. This is more efficient, but only works when a single server process is used. It is recommended to always leave this parameter with its default value of ``False``. Note: this method is a coroutine. """ await self.emit('message', data=data, to=to, room=room, skip_sid=skip_sid, namespace=namespace, callback=callback, ignore_queue=ignore_queue) async def call(self, event, data=None, to=None, sid=None, namespace=None, timeout=60, ignore_queue=False): """Emit a custom event to a client and wait for the response. This method issues an emit with a callback and waits for the callback to be invoked before returning. If the callback isn't invoked before the timeout, then a ``TimeoutError`` exception is raised. If the Socket.IO connection drops during the wait, this method still waits until the specified timeout. :param event: The event name. It can be any string. The event names ``'connect'``, ``'message'`` and ``'disconnect'`` are reserved and should not be used. :param data: The data to send to the client or clients. Data can be of type ``str``, ``bytes``, ``list`` or ``dict``. To send multiple arguments, use a tuple where each element is of one of the types indicated above. :param to: The session ID of the recipient client. :param sid: Alias for the ``to`` parameter. :param namespace: The Socket.IO namespace for the event. If this argument is omitted the event is emitted to the default namespace. :param timeout: The waiting timeout. If the timeout is reached before the client acknowledges the event, then a ``TimeoutError`` exception is raised. :param ignore_queue: Only used when a message queue is configured. If set to ``True``, the event is emitted to the client directly, without going through the queue. This is more efficient, but only works when a single server process is used. It is recommended to always leave this parameter with its default value of ``False``. Note: this method is not designed to be used concurrently. If multiple tasks are emitting at the same time to the same client connection, then messages composed of multiple packets may end up being sent in an incorrect sequence. Use standard concurrency solutions (such as a Lock object) to prevent this situation. Note 2: this method is a coroutine. """ if to is None and sid is None: raise ValueError('Cannot use call() to broadcast.') if not self.async_handlers: raise RuntimeError( 'Cannot use call() when async_handlers is False.') callback_event = self.eio.create_event() callback_args = [] def event_callback(*args): callback_args.append(args) callback_event.set() await self.emit(event, data=data, room=to or sid, namespace=namespace, callback=event_callback, ignore_queue=ignore_queue) try: await asyncio.wait_for(callback_event.wait(), timeout) except asyncio.TimeoutError: raise exceptions.TimeoutError() from None return callback_args[0] if len(callback_args[0]) > 1 \ else callback_args[0][0] if len(callback_args[0]) == 1 \ else None async def enter_room(self, sid, room, namespace=None): """Enter a room. This function adds the client to a room. The :func:`emit` and :func:`send` functions can optionally broadcast events to all the clients in a room. :param sid: Session ID of the client. :param room: Room name. If the room does not exist it is created. :param namespace: The Socket.IO namespace for the event. If this argument is omitted the default namespace is used. Note: this method is a coroutine. """ namespace = namespace or '/' self.logger.info('%s is entering room %s [%s]', sid, room, namespace) await self.manager.enter_room(sid, namespace, room) async def leave_room(self, sid, room, namespace=None): """Leave a room. This function removes the client from a room. :param sid: Session ID of the client. :param room: Room name. :param namespace: The Socket.IO namespace for the event. If this argument is omitted the default namespace is used. Note: this method is a coroutine. """ namespace = namespace or '/' self.logger.info('%s is leaving room %s [%s]', sid, room, namespace) await self.manager.leave_room(sid, namespace, room) async def close_room(self, room, namespace=None): """Close a room. This function removes all the clients from the given room. :param room: Room name. :param namespace: The Socket.IO namespace for the event. If this argument is omitted the default namespace is used. Note: this method is a coroutine. """ namespace = namespace or '/' self.logger.info('room %s is closing [%s]', room, namespace) await self.manager.close_room(room, namespace) async def get_session(self, sid, namespace=None): """Return the user session for a client. :param sid: The session id of the client. :param namespace: The Socket.IO namespace. If this argument is omitted the default namespace is used. The return value is a dictionary. Modifications made to this dictionary are not guaranteed to be preserved. If you want to modify the user session, use the ``session`` context manager instead. """ namespace = namespace or '/' eio_sid = self.manager.eio_sid_from_sid(sid, namespace) eio_session = await self.eio.get_session(eio_sid) return eio_session.setdefault(namespace, {}) async def save_session(self, sid, session, namespace=None): """Store the user session for a client. :param sid: The session id of the client. :param session: The session dictionary. :param namespace: The Socket.IO namespace. If this argument is omitted the default namespace is used. """ namespace = namespace or '/' eio_sid = self.manager.eio_sid_from_sid(sid, namespace) eio_session = await self.eio.get_session(eio_sid) eio_session[namespace] = session def session(self, sid, namespace=None): """Return the user session for a client with context manager syntax. :param sid: The session id of the client. This is a context manager that returns the user session dictionary for the client. Any changes that are made to this dictionary inside the context manager block are saved back to the session. Example usage:: @eio.on('connect') def on_connect(sid, environ): username = authenticate_user(environ) if not username: return False with eio.session(sid) as session: session['username'] = username @eio.on('message') def on_message(sid, msg): async with eio.session(sid) as session: print('received message from ', session['username']) """ class _session_context_manager(object): def __init__(self, server, sid, namespace): self.server = server self.sid = sid self.namespace = namespace self.session = None async def __aenter__(self): self.session = await self.server.get_session( sid, namespace=self.namespace) return self.session async def __aexit__(self, *args): await self.server.save_session(sid, self.session, namespace=self.namespace) return _session_context_manager(self, sid, namespace) async def disconnect(self, sid, namespace=None, ignore_queue=False): """Disconnect a client. :param sid: Session ID of the client. :param namespace: The Socket.IO namespace to disconnect. If this argument is omitted the default namespace is used. :param ignore_queue: Only used when a message queue is configured. If set to ``True``, the disconnect is processed locally, without broadcasting on the queue. It is recommended to always leave this parameter with its default value of ``False``. Note: this method is a coroutine. """ namespace = namespace or '/' if ignore_queue: delete_it = self.manager.is_connected(sid, namespace) else: delete_it = await self.manager.can_disconnect(sid, namespace) if delete_it: self.logger.info('Disconnecting %s [%s]', sid, namespace) eio_sid = self.manager.pre_disconnect(sid, namespace=namespace) await self._send_packet(eio_sid, self.packet_class( packet.DISCONNECT, namespace=namespace)) await self._trigger_event('disconnect', namespace, sid) await self.manager.disconnect(sid, namespace=namespace, ignore_queue=True) async def shutdown(self): """Stop Socket.IO background tasks. This method stops all background activity initiated by the Socket.IO server. It must be called before shutting down the web server. """ self.logger.info('Socket.IO is shutting down') await self.eio.shutdown() async def handle_request(self, *args, **kwargs): """Handle an HTTP request from the client. This is the entry point of the Socket.IO application. This function returns the HTTP response body to deliver to the client. Note: this method is a coroutine. """ return await self.eio.handle_request(*args, **kwargs) def start_background_task(self, target, *args, **kwargs): """Start a background task using the appropriate async model. This is a utility function that applications can use to start a background task using the method that is compatible with the selected async mode. :param target: the target function to execute. Must be a coroutine. :param args: arguments to pass to the function. :param kwargs: keyword arguments to pass to the function. The return value is a ``asyncio.Task`` object. """ return self.eio.start_background_task(target, *args, **kwargs) async def sleep(self, seconds=0): """Sleep for the requested amount of time using the appropriate async model. This is a utility function that applications can use to put a task to sleep without having to worry about using the correct call for the selected async mode. Note: this method is a coroutine. """ return await self.eio.sleep(seconds) def instrument(self, auth=None, mode='development', read_only=False, server_id=None, namespace='/admin', server_stats_interval=2): """Instrument the Socket.IO server for monitoring with the `Socket.IO Admin UI `_. :param auth: Authentication credentials for Admin UI access. Set to a dictionary with the expected login (usually ``username`` and ``password``) or a list of dictionaries if more than one set of credentials need to be available. For more complex authentication methods, set to a callable that receives the authentication dictionary as an argument and returns ``True`` if the user is allowed or ``False`` otherwise. To disable authentication, set this argument to ``False`` (not recommended, never do this on a production server). :param mode: The reporting mode. The default is ``'development'``, which is best used while debugging, as it may have a significant performance effect. Set to ``'production'`` to reduce the amount of information that is reported to the admin UI. :param read_only: If set to ``True``, the admin interface will be read-only, with no option to modify room assignments or disconnect clients. The default is ``False``. :param server_id: The server name to use for this server. If this argument is omitted, the server generates its own name. :param namespace: The Socket.IO namespace to use for the admin interface. The default is ``/admin``. :param server_stats_interval: The interval in seconds at which the server emits a summary of it stats to all connected admins. """ from .async_admin import InstrumentedAsyncServer return InstrumentedAsyncServer( self, auth=auth, mode=mode, read_only=read_only, server_id=server_id, namespace=namespace, server_stats_interval=server_stats_interval) async def _send_packet(self, eio_sid, pkt): """Send a Socket.IO packet to a client.""" encoded_packet = pkt.encode() if isinstance(encoded_packet, list): for ep in encoded_packet: await self.eio.send(eio_sid, ep) else: await self.eio.send(eio_sid, encoded_packet) async def _send_eio_packet(self, eio_sid, eio_pkt): """Send a raw Engine.IO packet to a client.""" await self.eio.send_packet(eio_sid, eio_pkt) async def _handle_connect(self, eio_sid, namespace, data): """Handle a client connection request.""" namespace = namespace or '/' sid = None if namespace in self.handlers or namespace in self.namespace_handlers \ or self.namespaces == '*' or namespace in self.namespaces: sid = await self.manager.connect(eio_sid, namespace) if sid is None: await self._send_packet(eio_sid, self.packet_class( packet.CONNECT_ERROR, data='Unable to connect', namespace=namespace)) return if self.always_connect: await self._send_packet(eio_sid, self.packet_class( packet.CONNECT, {'sid': sid}, namespace=namespace)) fail_reason = exceptions.ConnectionRefusedError().error_args try: if data: success = await self._trigger_event( 'connect', namespace, sid, self.environ[eio_sid], data) else: try: success = await self._trigger_event( 'connect', namespace, sid, self.environ[eio_sid]) except TypeError: success = await self._trigger_event( 'connect', namespace, sid, self.environ[eio_sid], None) except exceptions.ConnectionRefusedError as exc: fail_reason = exc.error_args success = False if success is False: if self.always_connect: self.manager.pre_disconnect(sid, namespace) await self._send_packet(eio_sid, self.packet_class( packet.DISCONNECT, data=fail_reason, namespace=namespace)) else: await self._send_packet(eio_sid, self.packet_class( packet.CONNECT_ERROR, data=fail_reason, namespace=namespace)) await self.manager.disconnect(sid, namespace, ignore_queue=True) elif not self.always_connect: await self._send_packet(eio_sid, self.packet_class( packet.CONNECT, {'sid': sid}, namespace=namespace)) async def _handle_disconnect(self, eio_sid, namespace): """Handle a client disconnect.""" namespace = namespace or '/' sid = self.manager.sid_from_eio_sid(eio_sid, namespace) if not self.manager.is_connected(sid, namespace): # pragma: no cover return self.manager.pre_disconnect(sid, namespace=namespace) await self._trigger_event('disconnect', namespace, sid) await self.manager.disconnect(sid, namespace, ignore_queue=True) async def _handle_event(self, eio_sid, namespace, id, data): """Handle an incoming client event.""" namespace = namespace or '/' sid = self.manager.sid_from_eio_sid(eio_sid, namespace) self.logger.info('received event "%s" from %s [%s]', data[0], sid, namespace) if not self.manager.is_connected(sid, namespace): self.logger.warning('%s is not connected to namespace %s', sid, namespace) return if self.async_handlers: task = self.start_background_task( self._handle_event_internal, self, sid, eio_sid, data, namespace, id) task_reference_holder.add(task) task.add_done_callback(task_reference_holder.discard) else: await self._handle_event_internal(self, sid, eio_sid, data, namespace, id) async def _handle_event_internal(self, server, sid, eio_sid, data, namespace, id): r = await server._trigger_event(data[0], namespace, sid, *data[1:]) if r != self.not_handled and id is not None: # send ACK packet with the response returned by the handler # tuples are expanded as multiple arguments if r is None: data = [] elif isinstance(r, tuple): data = list(r) else: data = [r] await server._send_packet(eio_sid, self.packet_class( packet.ACK, namespace=namespace, id=id, data=data)) async def _handle_ack(self, eio_sid, namespace, id, data): """Handle ACK packets from the client.""" namespace = namespace or '/' sid = self.manager.sid_from_eio_sid(eio_sid, namespace) self.logger.info('received ack from %s [%s]', sid, namespace) await self.manager.trigger_callback(sid, id, data) async def _trigger_event(self, event, namespace, *args): """Invoke an application event handler.""" # first see if we have an explicit handler for the event handler, args = self._get_event_handler(event, namespace, args) if handler: if asyncio.iscoroutinefunction(handler): try: ret = await handler(*args) except asyncio.CancelledError: # pragma: no cover ret = None else: ret = handler(*args) return ret # or else, forward the event to a namespace handler if one exists handler, args = self._get_namespace_handler(namespace, args) if handler: return await handler.trigger_event(event, *args) else: return self.not_handled async def _handle_eio_connect(self, eio_sid, environ): """Handle the Engine.IO connection event.""" if not self.manager_initialized: self.manager_initialized = True self.manager.initialize() self.environ[eio_sid] = environ async def _handle_eio_message(self, eio_sid, data): """Dispatch Engine.IO messages.""" if eio_sid in self._binary_packet: pkt = self._binary_packet[eio_sid] if pkt.add_attachment(data): del self._binary_packet[eio_sid] if pkt.packet_type == packet.BINARY_EVENT: await self._handle_event(eio_sid, pkt.namespace, pkt.id, pkt.data) else: await self._handle_ack(eio_sid, pkt.namespace, pkt.id, pkt.data) else: pkt = self.packet_class(encoded_packet=data) if pkt.packet_type == packet.CONNECT: await self._handle_connect(eio_sid, pkt.namespace, pkt.data) elif pkt.packet_type == packet.DISCONNECT: await self._handle_disconnect(eio_sid, pkt.namespace) elif pkt.packet_type == packet.EVENT: await self._handle_event(eio_sid, pkt.namespace, pkt.id, pkt.data) elif pkt.packet_type == packet.ACK: await self._handle_ack(eio_sid, pkt.namespace, pkt.id, pkt.data) elif pkt.packet_type == packet.BINARY_EVENT or \ pkt.packet_type == packet.BINARY_ACK: self._binary_packet[eio_sid] = pkt elif pkt.packet_type == packet.CONNECT_ERROR: raise ValueError('Unexpected CONNECT_ERROR packet.') else: raise ValueError('Unknown packet type.') async def _handle_eio_disconnect(self, eio_sid): """Handle Engine.IO disconnect event.""" for n in list(self.manager.get_namespaces()).copy(): await self._handle_disconnect(eio_sid, n) if eio_sid in self.environ: del self.environ[eio_sid] def _engineio_server_class(self): return engineio.AsyncServer