socketserver —- A framework for network servers

Source code:Lib/socketserver.py

The socketserver module simplifies the task of writing network servers.

There are four basic concrete server classes:

  • class socketserver.TCPServer(server_address, RequestHandlerClass, bind_and_activate=True)
  • This uses the Internet TCP protocol, which provides forcontinuous streams of data between the client and server.If bind_and_activate is true, the constructor automatically attempts toinvoke server_bind() andserver_activate(). The other parameters are passed tothe BaseServer base class.
  • class socketserver.UDPServer(server_address, RequestHandlerClass, bind_and_activate=True)
  • This uses datagrams, which are discrete packets of information that mayarrive out of order or be lost while in transit. The parameters arethe same as for TCPServer.
  • class socketserver.UnixStreamServer(server_address, RequestHandlerClass, bind_and_activate=True)
  • class socketserver.UnixDatagramServer(server_address, RequestHandlerClass, bind_and_activate=True)
  • These more infrequently used classes are similar to the TCP andUDP classes, but use Unix domain sockets; they're not available onnon-Unix platforms. The parameters are the same as forTCPServer.

These four classes process requests synchronously; each request must becompleted before the next request can be started. This isn't suitable if eachrequest takes a long time to complete, because it requires a lot of computation,or because it returns a lot of data which the client is slow to process. Thesolution is to create a separate process or thread to handle each request; theForkingMixIn and ThreadingMixIn mix-in classes can be used tosupport asynchronous behaviour.

Creating a server requires several steps. First, you must create a requesthandler class by subclassing the BaseRequestHandler class andoverriding its handle() method;this method will process incomingrequests. Second, you must instantiate one of the server classes, passing itthe server's address and the request handler class. It is recommended to usethe server in a with statement. Then call thehandle_request() orserve_forever() method of the server object toprocess one or many requests. Finally, call server_close()to close the socket (unless you used a with statement).

When inheriting from ThreadingMixIn for threaded connection behavior,you should explicitly declare how you want your threads to behave on an abruptshutdown. The ThreadingMixIn class defines an attributedaemon_threads, which indicates whether or not the server should wait forthread termination. You should set the flag explicitly if you would likethreads to behave autonomously; the default is False, meaning thatPython will not exit until all threads created by ThreadingMixIn haveexited.

Server classes have the same external methods and attributes, no matter whatnetwork protocol they use.

Server Creation Notes

There are five classes in an inheritance diagram, four of which representsynchronous servers of four types:

  1. +------------+
  2. | BaseServer |
  3. +------------+
  4.       |
  5.       v
  6. +-----------+        +------------------+
  7. | TCPServer |------->| UnixStreamServer |
  8. +-----------+        +------------------+
  9.       |
  10.       v
  11. +-----------+        +--------------------+
  12. | UDPServer |------->| UnixDatagramServer |
  13. +-----------+        +--------------------+

Note that UnixDatagramServer derives from UDPServer, not fromUnixStreamServer —- the only difference between an IP and a Unixstream server is the address family, which is simply repeated in both Unixserver classes.

  • class socketserver.ForkingMixIn
  • class socketserver.ThreadingMixIn
  • Forking and threading versions of each type of server can be createdusing these mix-in classes. For instance, ThreadingUDPServeris created as follows:
  1. class ThreadingUDPServer(ThreadingMixIn, UDPServer):
  2.     pass

The mix-in class comes first, since it overrides a method defined inUDPServer. Setting the various attributes also changes thebehavior of the underlying server mechanism.

ForkingMixIn and the Forking classes mentioned below areonly available on POSIX platforms that support fork().

socketserver.ForkingMixIn.server_close() waits until all childprocesses complete, except ifsocketserver.ForkingMixIn.block_on_close attribute is false.

socketserver.ThreadingMixIn.server_close() waits until all non-daemonthreads complete, except ifsocketserver.ThreadingMixIn.block_on_close attribute is false. Usedaemonic threads by settingThreadingMixIn.daemon_threads to True to not wait until threadscomplete.

在 3.7 版更改: socketserver.ForkingMixIn.server_close() andsocketserver.ThreadingMixIn.server_close() now waits until allchild processes and non-daemonic threads complete.Add a new socketserver.ForkingMixIn.block_on_close classattribute to opt-in for the pre-3.7 behaviour.

  • class socketserver.ForkingTCPServer
  • class socketserver.ForkingUDPServer
  • class socketserver.ThreadingTCPServer
  • class socketserver.ThreadingUDPServer
  • These classes are pre-defined using the mix-in classes.

To implement a service, you must derive a class from BaseRequestHandlerand redefine its handle() method.You can then run various versions ofthe service by combining one of the server classes with your request handlerclass. The request handler class must be different for datagram or streamservices. This can be hidden by using the handler subclassesStreamRequestHandler or DatagramRequestHandler.

Of course, you still have to use your head! For instance, it makes no sense touse a forking server if the service contains state in memory that can bemodified by different requests, since the modifications in the child processwould never reach the initial state kept in the parent process and passed toeach child. In this case, you can use a threading server, but you will probablyhave to use locks to protect the integrity of the shared data.

On the other hand, if you are building an HTTP server where all data is storedexternally (for instance, in the file system), a synchronous class willessentially render the service "deaf" while one request is being handled —which may be for a very long time if a client is slow to receive all the data ithas requested. Here a threading or forking server is appropriate.

In some cases, it may be appropriate to process part of a request synchronously,but to finish processing in a forked child depending on the request data. Thiscan be implemented by using a synchronous server and doing an explicit fork inthe request handler class handle() method.

Another approach to handling multiple simultaneous requests in an environmentthat supports neither threads nor fork() (or where these are tooexpensive or inappropriate for the service) is to maintain an explicit table ofpartially finished requests and to use selectors to decide whichrequest to work on next (or whether to handle a new incoming request). This isparticularly important for stream services where each client can potentially beconnected for a long time (if threads or subprocesses cannot be used). Seeasyncore for another way to manage this.

Server Objects

  • class socketserver.BaseServer(server_address, RequestHandlerClass)

  • This is the superclass of all Server objects in the module. It defines theinterface, given below, but does not implement most of the methods, which isdone in subclasses. The two parameters are stored in the respectiveserver_address and RequestHandlerClass attributes.

    • fileno()

    • Return an integer file descriptor for the socket on which the server islistening. This function is most commonly passed to selectors, toallow monitoring multiple servers in the same process.

    • handle_request()

    • Process a single request. This function calls the following methods inorder: get_request(), verify_request(), andprocess_request(). If the user-providedhandle() method of thehandler class raises an exception, the server's handle_error() methodwill be called. If no request is received within timeoutseconds, handle_timeout() will be called and handle_request()will return.

    • serveforever(_poll_interval=0.5)

    • Handle requests until an explicit shutdown() request. Poll forshutdown every poll_interval seconds.Ignores the timeout attribute. Italso calls service_actions(), which may be used by a subclass or mixinto provide actions specific to a given service. For example, theForkingMixIn class uses service_actions() to clean up zombiechild processes.

在 3.3 版更改: Added service_actions call to the serve_forever method.

  • service_actions()
  • This is called in the serve_forever() loop. This method can beoverridden by subclasses or mixin classes to perform actions specific toa given service, such as cleanup actions.

3.3 新版功能.

  • shutdown()

  • Tell the serve_forever() loop to stop and wait until it does.

  • server_close()

  • Clean up the server. May be overridden.

  • address_family

  • The family of protocols to which the server's socket belongs.Common examples are socket.AF_INET and socket.AF_UNIX.

  • RequestHandlerClass

  • The user-provided request handler class; an instance of this class is createdfor each request.

  • server_address

  • The address on which the server is listening. The format of addresses variesdepending on the protocol family;see the documentation for the socket modulefor details. For Internet protocols, this is a tuple containing a string givingthe address, and an integer port number: ('127.0.0.1', 80), for example.

  • socket

  • The socket object on which the server will listen for incoming requests.

The server classes support the following class variables:

  • allow_reuse_address

  • Whether the server will allow the reuse of an address. This defaults toFalse, and can be set in subclasses to change the policy.

  • request_queue_size

  • The size of the request queue. If it takes a long time to process a singlerequest, any requests that arrive while the server is busy are placed into aqueue, up to request_queue_size requests. Once the queue is full,further requests from clients will get a "Connection denied" error. The defaultvalue is usually 5, but this can be overridden by subclasses.

  • socket_type

  • The type of socket used by the server; socket.SOCK_STREAM andsocket.SOCK_DGRAM are two common values.

  • timeout

  • Timeout duration, measured in seconds, or None if no timeout isdesired. If handle_request() receives no incoming requests within thetimeout period, the handle_timeout() method is called.

There are various server methods that can be overridden by subclasses of baseserver classes like TCPServer; these methods aren't useful to externalusers of the server object.

  • finishrequest(_request, client_address)

  • Actually processes the request by instantiating RequestHandlerClass andcalling its handle() method.

  • get_request()

  • Must accept a request from the socket, and return a 2-tuple containing the _new_socket object to be used to communicate with the client, and the client'saddress.

  • handleerror(_request, client_address)

  • This function is called if the handle()method of a RequestHandlerClass instance raisesan exception. The default action is to print the traceback tostandard error and continue handling further requests.

在 3.6 版更改: Now only called for exceptions derived from the Exceptionclass.

  • handle_timeout()

  • This function is called when the timeout attribute has been set to avalue other than None and the timeout period has passed with norequests being received. The default action for forking servers isto collect the status of any child processes that have exited, whilein threading servers this method does nothing.

  • processrequest(_request, client_address)

  • Calls finish_request() to create an instance of theRequestHandlerClass. If desired, this function can create a new processor thread to handle the request; the ForkingMixIn andThreadingMixIn classes do this.

  • server_activate()

  • Called by the server's constructor to activate the server. The default behaviorfor a TCP server just invokes listen()on the server's socket. May be overridden.

  • server_bind()

  • Called by the server's constructor to bind the socket to the desired address.May be overridden.

  • verifyrequest(_request, client_address)

  • Must return a Boolean value; if the value is True, the request willbe processed, and if it's False, the request will be denied. Thisfunction can be overridden to implement access controls for a server. Thedefault implementation always returns True.

在 3.6 版更改: Support for the context manager protocol was added. Exiting thecontext manager is equivalent to calling server_close().

Request Handler Objects

  • class socketserver.BaseRequestHandler

  • This is the superclass of all request handler objects. It definesthe interface, given below. A concrete request handler subclass mustdefine a new handle() method, and can override any ofthe other methods. A new instance of the subclass is created for eachrequest.

    • setup()

    • Called before the handle() method to perform any initialization actionsrequired. The default implementation does nothing.

    • handle()

    • This function must do all the work required to service a request. Thedefault implementation does nothing. Several instance attributes areavailable to it; the request is available as self.request; the clientaddress as self.client_address; and the server instance asself.server, in case it needs access to per-server information.

The type of self.request is different for datagram or streamservices. For stream services, self.request is a socket object; fordatagram services, self.request is a pair of string and socket.

  • finish()
  • Called after the handle() method to perform any clean-up actionsrequired. The default implementation does nothing. If setup()raises an exception, this function will not be called.
  • class socketserver.StreamRequestHandler
  • class socketserver.DatagramRequestHandler
  • These BaseRequestHandler subclasses override thesetup() and finish()methods, and provide self.rfile and self.wfile attributes.The self.rfile and self.wfile attributes can beread or written, respectively, to get the request data or return datato the client.

The rfile attributes of both classes support theio.BufferedIOBase readable interface, andDatagramRequestHandler.wfile supports theio.BufferedIOBase writable interface.

在 3.6 版更改: StreamRequestHandler.wfile also supports theio.BufferedIOBase writable interface.

例子

socketserver.TCPServer Example

This is the server side:

  1. import socketserver
  2.  
  3. class MyTCPHandler(socketserver.BaseRequestHandler):
  4.     """
  5.     The request handler class for our server.
  6.  
  7.     It is instantiated once per connection to the server, and must
  8.     override the handle() method to implement communication to the
  9.     client.
  10.     """
  11.  
  12.     def handle(self):
  13.         # self.request is the TCP socket connected to the client
  14.         self.data = self.request.recv(1024).strip()
  15.         print("{} wrote:".format(self.client_address[0]))
  16.         print(self.data)
  17.         # just send back the same data, but upper-cased
  18.         self.request.sendall(self.data.upper())
  19.  
  20. if __name__ == "__main__":
  21.     HOST, PORT = "localhost", 9999
  22.  
  23.     # Create the server, binding to localhost on port 9999
  24.     with socketserver.TCPServer((HOST, PORT), MyTCPHandler) as server:
  25.         # Activate the server; this will keep running until you
  26.         # interrupt the program with Ctrl-C
  27.         server.serve_forever()

An alternative request handler class that makes use of streams (file-likeobjects that simplify communication by providing the standard file interface):

  1. class MyTCPHandler(socketserver.StreamRequestHandler):
  2.  
  3.     def handle(self):
  4.         # self.rfile is a file-like object created by the handler;
  5.         # we can now use e.g. readline() instead of raw recv() calls
  6.         self.data = self.rfile.readline().strip()
  7.         print("{} wrote:".format(self.client_address[0]))
  8.         print(self.data)
  9.         # Likewise, self.wfile is a file-like object used to write back
  10.         # to the client
  11.         self.wfile.write(self.data.upper())

The difference is that the readline() call in the second handler will callrecv() multiple times until it encounters a newline character, while thesingle recv() call in the first handler will just return what has been sentfrom the client in one sendall() call.

This is the client side:

  1. import socket
  2. import sys
  3.  
  4. HOST, PORT = "localhost", 9999
  5. data = " ".join(sys.argv[1:])
  6.  
  7. # Create a socket (SOCK_STREAM means a TCP socket)
  8. with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as sock:
  9.     # Connect to server and send data
  10.     sock.connect((HOST, PORT))
  11.     sock.sendall(bytes(data + "\n", "utf-8"))
  12.  
  13.     # Receive data from the server and shut down
  14.     received = str(sock.recv(1024), "utf-8")
  15.  
  16. print("Sent:     {}".format(data))
  17. print("Received: {}".format(received))

The output of the example should look something like this:

Server:

  1. $ python TCPServer.py
  2. 127.0.0.1 wrote:
  3. b'hello world with TCP'
  4. 127.0.0.1 wrote:
  5. b'python is nice'

Client:

  1. $ python TCPClient.py hello world with TCP
  2. Sent:     hello world with TCP
  3. Received: HELLO WORLD WITH TCP
  4. $ python TCPClient.py python is nice
  5. Sent:     python is nice
  6. Received: PYTHON IS NICE

socketserver.UDPServer Example

This is the server side:

  1. import socketserver
  2.  
  3. class MyUDPHandler(socketserver.BaseRequestHandler):
  4.     """
  5.     This class works similar to the TCP handler class, except that
  6.     self.request consists of a pair of data and client socket, and since
  7.     there is no connection the client address must be given explicitly
  8.     when sending data back via sendto().
  9.     """
  10.  
  11.     def handle(self):
  12.         data = self.request[0].strip()
  13.         socket = self.request[1]
  14.         print("{} wrote:".format(self.client_address[0]))
  15.         print(data)
  16.         socket.sendto(data.upper(), self.client_address)
  17.  
  18. if __name__ == "__main__":
  19.     HOST, PORT = "localhost", 9999
  20.     with socketserver.UDPServer((HOST, PORT), MyUDPHandler) as server:
  21.         server.serve_forever()

This is the client side:

  1. import socket
  2. import sys
  3.  
  4. HOST, PORT = "localhost", 9999
  5. data = " ".join(sys.argv[1:])
  6.  
  7. # SOCK_DGRAM is the socket type to use for UDP sockets
  8. sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
  9.  
  10. # As you can see, there is no connect() call; UDP has no connections.
  11. # Instead, data is directly sent to the recipient via sendto().
  12. sock.sendto(bytes(data + "\n", "utf-8"), (HOST, PORT))
  13. received = str(sock.recv(1024), "utf-8")
  14.  
  15. print("Sent:     {}".format(data))
  16. print("Received: {}".format(received))

The output of the example should look exactly like for the TCP server example.

Asynchronous Mixins

To build asynchronous handlers, use the ThreadingMixIn andForkingMixIn classes.

An example for the ThreadingMixIn class:

  1. import socket
  2. import threading
  3. import socketserver
  4.  
  5. class ThreadedTCPRequestHandler(socketserver.BaseRequestHandler):
  6.  
  7.     def handle(self):
  8.         data = str(self.request.recv(1024), 'ascii')
  9.         cur_thread = threading.current_thread()
  10.         response = bytes("{}: {}".format(cur_thread.name, data), 'ascii')
  11.         self.request.sendall(response)
  12.  
  13. class ThreadedTCPServer(socketserver.ThreadingMixIn, socketserver.TCPServer):
  14.     pass
  15.  
  16. def client(ip, port, message):
  17.     with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as sock:
  18.         sock.connect((ip, port))
  19.         sock.sendall(bytes(message, 'ascii'))
  20.         response = str(sock.recv(1024), 'ascii')
  21.         print("Received: {}".format(response))
  22.  
  23. if __name__ == "__main__":
  24.     # Port 0 means to select an arbitrary unused port
  25.     HOST, PORT = "localhost", 0
  26.  
  27.     server = ThreadedTCPServer((HOST, PORT), ThreadedTCPRequestHandler)
  28.     with server:
  29.         ip, port = server.server_address
  30.  
  31.         # Start a thread with the server -- that thread will then start one
  32.         # more thread for each request
  33.         server_thread = threading.Thread(target=server.serve_forever)
  34.         # Exit the server thread when the main thread terminates
  35.         server_thread.daemon = True
  36.         server_thread.start()
  37.         print("Server loop running in thread:", server_thread.name)
  38.  
  39.         client(ip, port, "Hello World 1")
  40.         client(ip, port, "Hello World 2")
  41.         client(ip, port, "Hello World 3")
  42.  
  43.         server.shutdown()

The output of the example should look something like this:

  1. $ python ThreadedTCPServer.py
  2. Server loop running in thread: Thread-1
  3. Received: Thread-2: Hello World 1
  4. Received: Thread-3: Hello World 2
  5. Received: Thread-4: Hello World 3

The ForkingMixIn class is used in the same way, except that the serverwill spawn a new process for each request.Available only on POSIX platforms that support fork().