Lib/SocketServer.py

   1 """Generic socket server classes.
   2
   3 This module tries to capture the various aspects of defining a server:
   4
   5 For socket-based servers:
   6
   7 - address family:
   8         - AF_INET{,6}: IP (Internet Protocol) sockets (default)
   9         - AF_UNIX: Unix domain sockets
  10         - others, e.g. AF_DECNET are conceivable (see <socket.h>
  11 - socket type:
  12         - SOCK_STREAM (reliable stream, e.g. TCP)
  13         - SOCK_DGRAM (datagrams, e.g. UDP)
  14
  15 For request-based servers (including socket-based):
  16
  17 - client address verification before further looking at the request
  18         (This is actually a hook for any processing that needs to look
  19          at the request before anything else, e.g. logging)
  20 - how to handle multiple requests:
  21         - synchronous (one request is handled at a time)
  22         - forking (each request is handled by a new process)
  23         - threading (each request is handled by a new thread)
  24
  25 The classes in this module favor the server type that is simplest to
  26 write: a synchronous TCP/IP server.  This is bad class design, but
  27 save some typing.  (There's also the issue that a deep class hierarchy
  28 slows down method lookups.)
  29
  30 There are five classes in an inheritance diagram, four of which represent
  31 synchronous servers of four types:
  32
  33         +------------+
  34         | BaseServer |
  35         +------------+
  36               |
  37               v
  38         +-----------+        +------------------+
  39         | TCPServer |------->| UnixStreamServer |
  40         +-----------+        +------------------+
  41               |
  42               v
  43         +-----------+        +--------------------+
  44         | UDPServer |------->| UnixDatagramServer |
  45         +-----------+        +--------------------+
  46
  47 Note that UnixDatagramServer derives from UDPServer, not from
  48 UnixStreamServer -- the only difference between an IP and a Unix
  49 stream server is the address family, which is simply repeated in both
  50 unix server classes.
  51
  52 Forking and threading versions of each type of server can be created
  53 using the ForkingServer and ThreadingServer mix-in classes.  For
  54 instance, a threading UDP server class is created as follows:
  55
  56         class ThreadingUDPServer(ThreadingMixIn, UDPServer): pass
  57
  58 The Mix-in class must come first, since it overrides a method defined
  59 in UDPServer!
  60
  61 To implement a service, you must derive a class from
  62 BaseRequestHandler and redefine its handle() method.  You can then run
  63 various versions of the service by combining one of the server classes
  64 with your request handler class.
  65
  66 The request handler class must be different for datagram or stream
  67 services.  This can be hidden by using the mix-in request handler
  68 classes StreamRequestHandler or DatagramRequestHandler.
  69
  70 Of course, you still have to use your head!
  71
  72 For instance, it makes no sense to use a forking server if the service
  73 contains state in memory that can be modified by requests (since the
  74 modifications in the child process would never reach the initial state
  75 kept in the parent process and passed to each child).  In this case,
  76 you can use a threading server, but you will probably have to use
  77 locks to avoid two requests that come in nearly simultaneous to apply
  78 conflicting changes to the server state.
  79
  80 On the other hand, if you are building e.g. an HTTP server, where all
  81 data is stored externally (e.g. in the file system), a synchronous
  82 class will essentially render the service "deaf" while one request is
  83 being handled -- which may be for a very long time if a client is slow
  84 to reqd all the data it has requested.  Here a threading or forking
  85 server is appropriate.
  86
  87 In some cases, it may be appropriate to process part of a request
  88 synchronously, but to finish processing in a forked child depending on
  89 the request data.  This can be implemented by using a synchronous
  90 server and doing an explicit fork in the request handler class
  91 handle() method.
  92
  93 Another approach to handling multiple simultaneous requests in an
  94 environment that supports neither threads nor fork (or where these are
  95 too expensive or inappropriate for the service) is to maintain an
  96 explicit table of partially finished requests and to use select() to
  97 decide which request to work on next (or whether to handle a new
  98 incoming request).  This is particularly important for stream services
  99 where each client can potentially be connected for a long time (if
 100 threads or subprocesses cannot be used).
 101
 102 Future work:
 103 - Standard classes for Sun RPC (which uses either UDP or TCP)
 104 - Standard mix-in classes to implement various authentication
 105   and encryption schemes
 106 - Standard framework for select-based multiplexing
 107
 108 XXX Open problems:
 109 - What to do with out-of-band data?
 110
 111 BaseServer:
 112 - split generic "request" functionality out into BaseServer class.
 113   Copyright (C) 2000  Luke Kenneth Casson Leighton <lkcl@samba.org>
 114
 115   example: read entries from a SQL database (requires overriding
 116   get_request() to return a table entry from the database).
 117   entry is processed by a RequestHandlerClass.
 118
 119 """
 120
 121 # Author of the BaseServer patch: Luke Kenneth Casson Leighton
 122
 123 # XXX Warning!
 124 # There is a test suite for this module, but it cannot be run by the
 125 # standard regression test.
 126 # To run it manually, run Lib/test/test_socketserver.py.
 127
 128 __version__ = "0.4"
 129
 130
 131 import socket
 132 import sys
 133 import os
 134
 135 __all__ = ["TCPServer","UDPServer","ForkingUDPServer","ForkingTCPServer",
 136            "ThreadingUDPServer","ThreadingTCPServer","BaseRequestHandler",
 137            "StreamRequestHandler","DatagramRequestHandler",
 138            "ThreadingMixIn", "ForkingMixIn"]
 139 if hasattr(socket, "AF_UNIX"):
 140     __all__.extend(["UnixStreamServer","UnixDatagramServer",
 141                     "ThreadingUnixStreamServer",
 142                     "ThreadingUnixDatagramServer"])
 143
 144 class BaseServer:
 145
 146     """Base class for server classes.
 147
 148     Methods for the caller:
 149
 150     - __init__(server_address, RequestHandlerClass)
 151     - serve_forever()
 152     - handle_request()  # if you do not use serve_forever()
 153     - fileno() -> int   # for select()
 154
 155     Methods that may be overridden:
 156
 157     - server_bind()
 158     - server_activate()
 159     - get_request() -> request, client_address
 160     - verify_request(request, client_address)
 161     - server_close()
 162     - process_request(request, client_address)
 163     - close_request(request)
 164     - handle_error()
 165
 166     Methods for derived classes:
 167
 168     - finish_request(request, client_address)
 169
 170     Class variables that may be overridden by derived classes or
 171     instances:
 172
 173     - address_family
 174     - socket_type
 175     - reuse_address
 176
 177     Instance variables:
 178
 179     - RequestHandlerClass
 180     - socket
 181
 182     """
 183
 184     def __init__(self, server_address, RequestHandlerClass):
 185         """Constructor.  May be extended, do not override."""
 186         self.server_address = server_address
 187         self.RequestHandlerClass = RequestHandlerClass
 188
 189     def server_activate(self):
 190         """Called by constructor to activate the server.
 191
 192         May be overridden.
 193
 194         """
 195         pass
 196
 197     def serve_forever(self):
 198         """Handle one request at a time until doomsday."""
 199         while 1:
 200             self.handle_request()
 201
 202     # The distinction between handling, getting, processing and
 203     # finishing a request is fairly arbitrary.  Remember:
 204     #
 205     # - handle_request() is the top-level call.  It calls
 206     #   get_request(), verify_request() and process_request()
 207     # - get_request() is different for stream or datagram sockets
 208     # - process_request() is the place that may fork a new process
 209     #   or create a new thread to finish the request
 210     # - finish_request() instantiates the request handler class;
 211     #   this constructor will handle the request all by itself
 212
 213     def handle_request(self):
 214         """Handle one request, possibly blocking."""
 215         try:
 216             request, client_address = self.get_request()
 217         except socket.error:
 218             return
 219         if self.verify_request(request, client_address):
 220             try:
 221                 self.process_request(request, client_address)
 222             except:
 223                 self.handle_error(request, client_address)
 224                 self.close_request(request)
 225
 226     def verify_request(self, request, client_address):
 227         """Verify the request.  May be overridden.
 228
 229         Return true if we should proceed with this request.
 230
 231         """
 232         return 1
 233
 234     def process_request(self, request, client_address):
 235         """Call finish_request.
 236
 237         Overridden by ForkingMixIn and ThreadingMixIn.
 238
 239         """
 240         self.finish_request(request, client_address)
 241         self.close_request(request)
 242
 243     def server_close(self):
 244         """Called to clean-up the server.
 245
 246         May be overridden.
 247
 248         """
 249         pass
 250
 251     def finish_request(self, request, client_address):
 252         """Finish one request by instantiating RequestHandlerClass."""
 253         self.RequestHandlerClass(request, client_address, self)
 254
 255     def close_request(self, request):
 256         """Called to clean up an individual request."""
 257         pass
 258
 259     def handle_error(self, request, client_address):
 260         """Handle an error gracefully.  May be overridden.
 261
 262         The default is to print a traceback and continue.
 263
 264         """
 265         print '-'*40
 266         print 'Exception happened during processing of request from',
 267         print client_address
 268         import traceback
 269         traceback.print_exc() # XXX But this goes to stderr!
 270         print '-'*40
 271
 272
 273 class TCPServer(BaseServer):
 274
 275     """Base class for various socket-based server classes.
 276
 277     Defaults to synchronous IP stream (i.e., TCP).
 278
 279     Methods for the caller:
 280
 281     - __init__(server_address, RequestHandlerClass)
 282     - serve_forever()
 283     - handle_request()  # if you don't use serve_forever()
 284     - fileno() -> int   # for select()
 285
 286     Methods that may be overridden:
 287
 288     - server_bind()
 289     - server_activate()
 290     - get_request() -> request, client_address
 291     - verify_request(request, client_address)
 292     - process_request(request, client_address)
 293     - close_request(request)
 294     - handle_error()
 295
 296     Methods for derived classes:
 297
 298     - finish_request(request, client_address)
 299
 300     Class variables that may be overridden by derived classes or
 301     instances:
 302
 303     - address_family
 304     - socket_type
 305     - request_queue_size (only for stream sockets)
 306     - reuse_address
 307
 308     Instance variables:
 309
 310     - server_address
 311     - RequestHandlerClass
 312     - socket
 313
 314     """
 315
 316     address_family = socket.AF_INET
 317
 318     socket_type = socket.SOCK_STREAM
 319
 320     request_queue_size = 5
 321
 322     allow_reuse_address = 0
 323
 324     def __init__(self, server_address, RequestHandlerClass):
 325         """Constructor.  May be extended, do not override."""
 326         BaseServer.__init__(self, server_address, RequestHandlerClass)
 327         self.socket = socket.socket(self.address_family,
 328                                     self.socket_type)
 329         self.server_bind()
 330         self.server_activate()
 331
 332     def server_bind(self):
 333         """Called by constructor to bind the socket.
 334
 335         May be overridden.
 336
 337         """
 338         if self.allow_reuse_address:
 339             self.socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
 340         self.socket.bind(self.server_address)
 341
 342     def server_activate(self):
 343         """Called by constructor to activate the server.
 344
 345         May be overridden.
 346
 347         """
 348         self.socket.listen(self.request_queue_size)
 349
 350     def server_close(self):
 351         """Called to clean-up the server.
 352
 353         May be overridden.
 354
 355         """
 356         self.socket.close()
 357
 358     def fileno(self):
 359         """Return socket file number.
 360
 361         Interface required by select().
 362
 363         """
 364         return self.socket.fileno()
 365
 366     def get_request(self):
 367         """Get the request and client address from the socket.
 368
 369         May be overridden.
 370
 371         """
 372         return self.socket.accept()
 373
 374     def close_request(self, request):
 375         """Called to clean up an individual request."""
 376         request.close()
 377
 378
 379 class UDPServer(TCPServer):
 380
 381     """UDP server class."""
 382
 383     allow_reuse_address = 0
 384
 385     socket_type = socket.SOCK_DGRAM
 386
 387     max_packet_size = 8192
 388
 389     def get_request(self):
 390         data, client_addr = self.socket.recvfrom(self.max_packet_size)
 391         return (data, self.socket), client_addr
 392
 393     def server_activate(self):
 394         # No need to call listen() for UDP.
 395         pass
 396
 397     def close_request(self, request):
 398         # No need to close anything.
 399         pass
 400
 401 class ForkingMixIn:
 402
 403     """Mix-in class to handle each request in a new process."""
 404
 405     active_children = None
 406     max_children = 40
 407
 408     def collect_children(self):
 409         """Internal routine to wait for died children."""
 410         while self.active_children:
 411             if len(self.active_children) < self.max_children:
 412                 options = os.WNOHANG
 413             else:
 414                 # If the maximum number of children are already
 415                 # running, block while waiting for a child to exit
 416                 options = 0
 417             try:
 418                 pid, status = os.waitpid(0, options)
 419             except os.error:
 420                 pid = None
 421             if not pid: break
 422             self.active_children.remove(pid)
 423
 424     def process_request(self, request, client_address):
 425         """Fork a new subprocess to process the request."""
 426         self.collect_children()
 427         pid = os.fork()
 428         if pid:
 429             # Parent process
 430             if self.active_children is None:
 431                 self.active_children = []
 432             self.active_children.append(pid)
 433             self.close_request(request)
 434             return
 435         else:
 436             # Child process.
 437             # This must never return, hence os._exit()!
 438             try:
 439                 self.finish_request(request, client_address)
 440                 os._exit(0)
 441             except:
 442                 try:
 443                     self.handle_error(request, client_address)
 444                 finally:
 445                     os._exit(1)
 446
 447
 448 class ThreadingMixIn:
 449     """Mix-in class to handle each request in a new thread."""
 450
 451     def process_request(self, request, client_address):
 452         """Start a new thread to process the request."""
 453         import threading
 454         t = threading.Thread(target = self.finish_request,
 455                              args = (request, client_address))
 456         t.start()
 457
 458
 459 class ForkingUDPServer(ForkingMixIn, UDPServer): pass
 460 class ForkingTCPServer(ForkingMixIn, TCPServer): pass
 461
 462 class ThreadingUDPServer(ThreadingMixIn, UDPServer): pass
 463 class ThreadingTCPServer(ThreadingMixIn, TCPServer): pass
 464
 465 if hasattr(socket, 'AF_UNIX'):
 466
 467     class UnixStreamServer(TCPServer):
 468         address_family = socket.AF_UNIX
 469
 470     class UnixDatagramServer(UDPServer):
 471         address_family = socket.AF_UNIX
 472
 473     class ThreadingUnixStreamServer(ThreadingMixIn, UnixStreamServer): pass
 474
 475     class ThreadingUnixDatagramServer(ThreadingMixIn, UnixDatagramServer): pass
 476
 477 class BaseRequestHandler:
 478
 479     """Base class for request handler classes.
 480
 481     This class is instantiated for each request to be handled.  The
 482     constructor sets the instance variables request, client_address
 483     and server, and then calls the handle() method.  To implement a
 484     specific service, all you need to do is to derive a class which
 485     defines a handle() method.
 486
 487     The handle() method can find the request as self.request, the
 488     client address as self.client_address, and the server (in case it
 489     needs access to per-server information) as self.server.  Since a
 490     separate instance is created for each request, the handle() method
 491     can define arbitrary other instance variariables.
 492
 493     """
 494
 495     def __init__(self, request, client_address, server):
 496         self.request = request
 497         self.client_address = client_address
 498         self.server = server
 499         try:
 500             self.setup()
 501             self.handle()
 502             self.finish()
 503         finally:
 504             sys.exc_traceback = None    # Help garbage collection
 505
 506     def setup(self):
 507         pass
 508
 509     def __del__(self):
 510         pass
 511
 512     def handle(self):
 513         pass
 514
 515     def finish(self):
 516         pass
 517
 518
 519 # The following two classes make it possible to use the same service
 520 # class for stream or datagram servers.
 521 # Each class sets up these instance variables:
 522 # - rfile: a file object from which receives the request is read
 523 # - wfile: a file object to which the reply is written
 524 # When the handle() method returns, wfile is flushed properly
 525
 526
 527 class StreamRequestHandler(BaseRequestHandler):
 528
 529     """Define self.rfile and self.wfile for stream sockets."""
 530
 531     # Default buffer sizes for rfile, wfile.
 532     # We default rfile to buffered because otherwise it could be
 533     # really slow for large data (a getc() call per byte); we make
 534     # wfile unbuffered because (a) often after a write() we want to
 535     # read and we need to flush the line; (b) big writes to unbuffered
 536     # files are typically optimized by stdio even when big reads
 537     # aren't.
 538     rbufsize = -1
 539     wbufsize = 0
 540
 541     def setup(self):
 542         self.connection = self.request
 543         self.rfile = self.connection.makefile('rb', self.rbufsize)
 544         self.wfile = self.connection.makefile('wb', self.wbufsize)
 545
 546     def finish(self):
 547         self.wfile.flush()
 548         self.wfile.close()
 549         self.rfile.close()
 550
 551
 552 class DatagramRequestHandler(BaseRequestHandler):
 553
 554     # XXX Regrettably, I cannot get this working on Linux;
 555     # s.recvfrom() doesn't return a meaningful client address.
 556
 557     """Define self.rfile and self.wfile for datagram sockets."""
 558
 559     def setup(self):
 560         import StringIO
 561         self.packet, self.socket = self.request
 562         self.rfile = StringIO.StringIO(self.packet)
 563         self.wfile = StringIO.StringIO(self.packet)
 564
 565     def finish(self):
 566         self.socket.sendto(self.wfile.getvalue(), self.client_address)