"""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 ForkingMixIn and ThreadingMixIn 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! Setting the various member variables also changes 60 the behavior of the underlying server mechanism. 61 62 To implement a service, you must derive a class from 63 BaseRequestHandler and redefine its handle() method. You can then run 64 various versions of the service by combining one of the server classes 65 with your request handler class. 66 67 The request handler class must be different for datagram or stream 68 services. This can be hidden by using the request handler 69 subclasses StreamRequestHandler or DatagramRequestHandler. 70 71 Of course, you still have to use your head! 72 73 For instance, it makes no sense to use a forking server if the service 74 contains state in memory that can be modified by requests (since the 75 modifications in the child process would never reach the initial state 76 kept in the parent process and passed to each child). In this case, 77 you can use a threading server, but you will probably have to use 78 locks to avoid two requests that come in nearly simultaneous to apply 79 conflicting changes to the server state. 80 81 On the other hand, if you are building e.g. an HTTP server, where all 82 data is stored externally (e.g. in the file system), a synchronous 83 class will essentially render the service "deaf" while one request is 84 being handled -- which may be for a very long time if a client is slow 85 to read all the data it has requested. Here a threading or forking 86 server is appropriate. 87 88 In some cases, it may be appropriate to process part of a request 89 synchronously, but to finish processing in a forked child depending on 90 the request data. This can be implemented by using a synchronous 91 server and doing an explicit fork in the request handler class 92 handle() method. 93 94 Another approach to handling multiple simultaneous requests in an 95 environment that supports neither threads nor fork (or where these are 96 too expensive or inappropriate for the service) is to maintain an 97 explicit table of partially finished requests and to use select() to 98 decide which request to work on next (or whether to handle a new 99 incoming request). This is particularly important for stream services 100 where each client can potentially be connected for a long time (if 101 threads or subprocesses cannot be used). 102 103 Future work: 104 - Standard classes for Sun RPC (which uses either UDP or TCP) 105 - Standard mix-in classes to implement various authentication 106 and encryption schemes 107 - Standard framework for select-based multiplexing 108 109 XXX Open problems: 110 - What to do with out-of-band data? 111 112 BaseServer: 113 - split generic "request" functionality out into BaseServer class. 114 Copyright (C) 2000 Luke Kenneth Casson Leighton <lkcl@samba.org> 115 116 example: read entries from a SQL database (requires overriding 117 get_request() to return a table entry from the database). 118 entry is processed by a RequestHandlerClass. 119 120 """ 121 122 # Author of the BaseServer patch: Luke Kenneth Casson Leighton 123 124 # XXX Warning! 125 # There is a test suite for this module, but it cannot be run by the 126 # standard regression test. 127 # To run it manually, run Lib/test/test_socketserver.py. 128 129 __version__ = "0.4" 130 131 132 import socket 133 import select 134 import sys 135 import os 136 import errno 137 try: 138 import threading 139 except ImportError: 140 import dummy_threading as threading 141 142 __all__ = ["TCPServer","UDPServer","ForkingUDPServer","ForkingTCPServer", 143 "ThreadingUDPServer","ThreadingTCPServer","BaseRequestHandler", 144 "StreamRequestHandler","DatagramRequestHandler", 145 "ThreadingMixIn", "ForkingMixIn"] 146 if hasattr(socket, "AF_UNIX"): 147 __all__.extend(["UnixStreamServer","UnixDatagramServer", 148 "ThreadingUnixStreamServer", 149 "ThreadingUnixDatagramServer"]) 150 151 def _eintr_retry(func, *args): 152 """restart a system call interrupted by EINTR""" 153 while True: 154 try: 155 return func(*args) 156 except (OSError, select.error) as e: 157 if e.args[0] != errno.EINTR: 158 raise 159 160 class BaseServer: 161 162 """Base class for server classes. 163 164 Methods for the caller: 165 166 - __init__(server_address, RequestHandlerClass) 167 - serve_forever(poll_interval=0.5) 168 - shutdown() 169 - handle_request() # if you do not use serve_forever() 170 - fileno() -> int # for select() 171 172 Methods that may be overridden: 173 174 - server_bind() 175 - server_activate() 176 - get_request() -> request, client_address 177 - handle_timeout() 178 - verify_request(request, client_address) 179 - server_close() 180 - process_request(request, client_address) 181 - shutdown_request(request) 182 - close_request(request) 183 - handle_error() 184 185 Methods for derived classes: 186 187 - finish_request(request, client_address) 188 189 Class variables that may be overridden by derived classes or 190 instances: 191 192 - timeout 193 - address_family 194 - socket_type 195 - allow_reuse_address 196 197 Instance variables: 198 199 - RequestHandlerClass 200 - socket 201 202 """ 203 204 timeout = None 205 206 def __init__(self, server_address, RequestHandlerClass): 207 """Constructor. May be extended, do not override.""" 208 self.server_address = server_address 209 self.RequestHandlerClass = RequestHandlerClass 210 self.__is_shut_down = threading.Event() 211 self.__shutdown_request = False 212 213 def server_activate(self): 214 """Called by constructor to activate the server. 215 216 May be overridden. 217 218 """ 219 pass 220 221 def serve_forever(self, poll_interval=0.5): 222 """Handle one request at a time until shutdown. 223 224 Polls for shutdown every poll_interval seconds. Ignores 225 self.timeout. If you need to do periodic tasks, do them in 226 another thread. 227 """ 228 self.__is_shut_down.clear() 229 try: 230 while not self.__shutdown_request: 231 # XXX: Consider using another file descriptor or 232 # connecting to the socket to wake this up instead of 233 # polling. Polling reduces our responsiveness to a 234 # shutdown request and wastes cpu at all other times. 235 r, w, e = _eintr_retry(select.select, [self], [], [], 236 poll_interval) 237 if self in r: 238 self._handle_request_noblock() 239 finally: 240 self.__shutdown_request = False 241 self.__is_shut_down.set() 242 243 def shutdown(self): 244 """Stops the serve_forever loop. 245 246 Blocks until the loop has finished. This must be called while 247 serve_forever() is running in another thread, or it will 248 deadlock. 249 """ 250 self.__shutdown_request = True 251 self.__is_shut_down.wait() 252 253 # The distinction between handling, getting, processing and 254 # finishing a request is fairly arbitrary. Remember: 255 # 256 # - handle_request() is the top-level call. It calls 257 # select, get_request(), verify_request() and process_request() 258 # - get_request() is different for stream or datagram sockets 259 # - process_request() is the place that may fork a new process 260 # or create a new thread to finish the request 261 # - finish_request() instantiates the request handler class; 262 # this constructor will handle the request all by itself 263 264 def handle_request(self): 265 """Handle one request, possibly blocking. 266 267 Respects self.timeout. 268 """ 269 # Support people who used socket.settimeout() to escape 270 # handle_request before self.timeout was available. 271 timeout = self.socket.gettimeout() 272 if timeout is None: 273 timeout = self.timeout 274 elif self.timeout is not None: 275 timeout = min(timeout, self.timeout) 276 fd_sets = _eintr_retry(select.select, [self], [], [], timeout) 277 if not fd_sets[0]: 278 self.handle_timeout() 279 return 280 self._handle_request_noblock() 281 282 def _handle_request_noblock(self): 283 """Handle one request, without blocking. 284 285 I assume that select.select has returned that the socket is 286 readable before this function was called, so there should be 287 no risk of blocking in get_request(). 288 """ 289 try: 290 request, client_address = self.get_request() 291 except socket.error: 292 return 293 if self.verify_request(request, client_address): 294 try: 295 self.process_request(request, client_address) 296 except: 297 self.handle_error(request, client_address) 298 self.shutdown_request(request) 299 300 def handle_timeout(self): 301 """Called if no new request arrives within self.timeout. 302 303 Overridden by ForkingMixIn. 304 """ 305 pass 306 307 def verify_request(self, request, client_address): 308 """Verify the request. May be overridden. 309 310 Return True if we should proceed with this request. 311 312 """ 313 return True 314 315 def process_request(self, request, client_address): 316 """Call finish_request. 317 318 Overridden by ForkingMixIn and ThreadingMixIn. 319 320 """ 321 self.finish_request(request, client_address) 322 self.shutdown_request(request) 323 324 def server_close(self): 325 """Called to clean-up the server. 326 327 May be overridden. 328 329 """ 330 pass 331 332 def finish_request(self, request, client_address): 333 """Finish one request by instantiating RequestHandlerClass.""" 334 self.RequestHandlerClass(request, client_address, self) 335 336 def shutdown_request(self, request): 337 """Called to shutdown and close an individual request.""" 338 self.close_request(request) 339 340 def close_request(self, request): 341 """Called to clean up an individual request.""" 342 pass 343 344 def handle_error(self, request, client_address): 345 """Handle an error gracefully. May be overridden. 346 347 The default is to print a traceback and continue. 348 349 """ 350 print '-'*40 351 print 'Exception happened during processing of request from', 352 print client_address 353 import traceback 354 traceback.print_exc() # XXX But this goes to stderr! 355 print '-'*40 356 357 358 class TCPServer(BaseServer): 359 360 """Base class for various socket-based server classes. 361 362 Defaults to synchronous IP stream (i.e., TCP). 363 364 Methods for the caller: 365 366 - __init__(server_address, RequestHandlerClass, bind_and_activate=True) 367 - serve_forever(poll_interval=0.5) 368 - shutdown() 369 - handle_request() # if you don't use serve_forever() 370 - fileno() -> int # for select() 371 372 Methods that may be overridden: 373 374 - server_bind() 375 - server_activate() 376 - get_request() -> request, client_address 377 - handle_timeout() 378 - verify_request(request, client_address) 379 - process_request(request, client_address) 380 - shutdown_request(request) 381 - close_request(request) 382 - handle_error() 383 384 Methods for derived classes: 385 386 - finish_request(request, client_address) 387 388 Class variables that may be overridden by derived classes or 389 instances: 390 391 - timeout 392 - address_family 393 - socket_type 394 - request_queue_size (only for stream sockets) 395 - allow_reuse_address 396 397 Instance variables: 398 399 - server_address 400 - RequestHandlerClass 401 - socket 402 403 """ 404 405 address_family = socket.AF_INET 406 407 socket_type = socket.SOCK_STREAM 408 409 request_queue_size = 5 410 411 allow_reuse_address = False 412 413 def __init__(self, server_address, RequestHandlerClass, bind_and_activate=True): 414 """Constructor. May be extended, do not override.""" 415 BaseServer.__init__(self, server_address, RequestHandlerClass) 416 self.socket = socket.socket(self.address_family, 417 self.socket_type) 418 if bind_and_activate: 419 self.server_bind() 420 self.server_activate() 421 422 def server_bind(self): 423 """Called by constructor to bind the socket. 424 425 May be overridden. 426 427 """ 428 if self.allow_reuse_address: 429 self.socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1) 430 self.socket.bind(self.server_address) 431 self.server_address = self.socket.getsockname() 432 433 def server_activate(self): 434 """Called by constructor to activate the server. 435 436 May be overridden. 437 438 """ 439 self.socket.listen(self.request_queue_size) 440 441 def server_close(self): 442 """Called to clean-up the server. 443 444 May be overridden. 445 446 """ 447 self.socket.close() 448 449 def fileno(self): 450 """Return socket file number. 451 452 Interface required by select(). 453 454 """ 455 return self.socket.fileno() 456 457 def get_request(self): 458 """Get the request and client address from the socket. 459 460 May be overridden. 461 462 """ 463 return self.socket.accept() 464 465 def shutdown_request(self, request): 466 """Called to shutdown and close an individual request.""" 467 try: 468 #explicitly shutdown. socket.close() merely releases 469 #the socket and waits for GC to perform the actual close. 470 request.shutdown(socket.SHUT_WR) 471 except socket.error: 472 pass #some platforms may raise ENOTCONN here 473 self.close_request(request) 474 475 def close_request(self, request): 476 """Called to clean up an individual request.""" 477 request.close() 478 479 480 class UDPServer(TCPServer): 481 482 """UDP server class.""" 483 484 allow_reuse_address = False 485 486 socket_type = socket.SOCK_DGRAM 487 488 max_packet_size = 8192 489 490 def get_request(self): 491 data, client_addr = self.socket.recvfrom(self.max_packet_size) 492 return (data, self.socket), client_addr 493 494 def server_activate(self): 495 # No need to call listen() for UDP. 496 pass 497 498 def shutdown_request(self, request): 499 # No need to shutdown anything. 500 self.close_request(request) 501 502 def close_request(self, request): 503 # No need to close anything. 504 pass 505 506 class ForkingMixIn: 507 508 """Mix-in class to handle each request in a new process.""" 509 510 timeout = 300 511 active_children = None 512 max_children = 40 513 514 def collect_children(self): 515 """Internal routine to wait for children that have exited.""" 516 if self.active_children is None: return 517 while len(self.active_children) >= self.max_children: 518 # XXX: This will wait for any child process, not just ones 519 # spawned by this library. This could confuse other 520 # libraries that expect to be able to wait for their own 521 # children. 522 try: 523 pid, status = os.waitpid(0, 0) 524 except os.error: 525 pid = None 526 if pid not in self.active_children: continue 527 self.active_children.remove(pid) 528 529 # XXX: This loop runs more system calls than it ought 530 # to. There should be a way to put the active_children into a 531 # process group and then use os.waitpid(-pgid) to wait for any 532 # of that set, but I couldn't find a way to allocate pgids 533 # that couldn't collide. 534 for child in self.active_children: 535 try: 536 pid, status = os.waitpid(child, os.WNOHANG) 537 except os.error: 538 pid = None 539 if not pid: continue 540 try: 541 self.active_children.remove(pid) 542 except ValueError, e: 543 raise ValueError('%s. x=%d and list=%r' % (e.message, pid, 544 self.active_children)) 545 546 def handle_timeout(self): 547 """Wait for zombies after self.timeout seconds of inactivity. 548 549 May be extended, do not override. 550 """ 551 self.collect_children() 552 553 def process_request(self, request, client_address): 554 """Fork a new subprocess to process the request.""" 555 self.collect_children() 556 pid = os.fork() 557 if pid: 558 # Parent process 559 if self.active_children is None: 560 self.active_children = [] 561 self.active_children.append(pid) 562 self.close_request(request) #close handle in parent process 563 return 564 else: 565 # Child process. 566 # This must never return, hence os._exit()! 567 try: 568 self.finish_request(request, client_address) 569 self.shutdown_request(request) 570 os._exit(0) 571 except: 572 try: 573 self.handle_error(request, client_address) 574 self.shutdown_request(request) 575 finally: 576 os._exit(1) 577 578 579 class ThreadingMixIn: 580 """Mix-in class to handle each request in a new thread.""" 581 582 # Decides how threads will act upon termination of the 583 # main process 584 daemon_threads = False 585 586 def process_request_thread(self, request, client_address): 587 """Same as in BaseServer but as a thread. 588 589 In addition, exception handling is done here. 590 591 """ 592 try: 593 self.finish_request(request, client_address) 594 self.shutdown_request(request) 595 except: 596 self.handle_error(request, client_address) 597 self.shutdown_request(request) 598 599 def process_request(self, request, client_address): 600 """Start a new thread to process the request.""" 601 t = threading.Thread(target = self.process_request_thread, 602 args = (request, client_address)) 603 t.daemon = self.daemon_threads 604 t.start() 605 606 607 class ForkingUDPServer(ForkingMixIn, UDPServer): pass 608 class ForkingTCPServer(ForkingMixIn, TCPServer): pass 609 610 class ThreadingUDPServer(ThreadingMixIn, UDPServer): pass 611 class ThreadingTCPServer(ThreadingMixIn, TCPServer): pass 612 613 if hasattr(socket, 'AF_UNIX'): 614 615 class UnixStreamServer(TCPServer): 616 address_family = socket.AF_UNIX 617 618 class UnixDatagramServer(UDPServer): 619 address_family = socket.AF_UNIX 620 621 class ThreadingUnixStreamServer(ThreadingMixIn, UnixStreamServer): pass 622 623 class ThreadingUnixDatagramServer(ThreadingMixIn, UnixDatagramServer): pass 624 625 class BaseRequestHandler: 626 627 """Base class for request handler classes. 628 629 This class is instantiated for each request to be handled. The 630 constructor sets the instance variables request, client_address 631 and server, and then calls the handle() method. To implement a 632 specific service, all you need to do is to derive a class which 633 defines a handle() method. 634 635 The handle() method can find the request as self.request, the 636 client address as self.client_address, and the server (in case it 637 needs access to per-server information) as self.server. Since a 638 separate instance is created for each request, the handle() method 639 can define arbitrary other instance variariables. 640 641 """ 642 643 def __init__(self, request, client_address, server): 644 self.request = request 645 self.client_address = client_address 646 self.server = server 647 self.setup() 648 try: 649 self.handle() 650 finally: 651 self.finish() 652 653 def setup(self): 654 pass 655 656 def handle(self): 657 pass 658 659 def finish(self): 660 pass 661 662 663 # The following two classes make it possible to use the same service 664 # class for stream or datagram servers. 665 # Each class sets up these instance variables: 666 # - rfile: a file object from which receives the request is read 667 # - wfile: a file object to which the reply is written 668 # When the handle() method returns, wfile is flushed properly 669 670 671 class StreamRequestHandler(BaseRequestHandler): 672 673 """Define self.rfile and self.wfile for stream sockets.""" 674 675 # Default buffer sizes for rfile, wfile. 676 # We default rfile to buffered because otherwise it could be 677 # really slow for large data (a getc() call per byte); we make 678 # wfile unbuffered because (a) often after a write() we want to 679 # read and we need to flush the line; (b) big writes to unbuffered 680 # files are typically optimized by stdio even when big reads 681 # aren't. 682 rbufsize = -1 683 wbufsize = 0 684 685 # A timeout to apply to the request socket, if not None. 686 timeout = None 687 688 # Disable nagle algorithm for this socket, if True. 689 # Use only when wbufsize != 0, to avoid small packets. 690 disable_nagle_algorithm = False 691 692 def setup(self): 693 self.connection = self.request 694 if self.timeout is not None: 695 self.connection.settimeout(self.timeout) 696 if self.disable_nagle_algorithm: 697 self.connection.setsockopt(socket.IPPROTO_TCP, 698 socket.TCP_NODELAY, True) 699 self.rfile = self.connection.makefile('rb', self.rbufsize) 700 self.wfile = self.connection.makefile('wb', self.wbufsize) 701 702 def finish(self): 703 if not self.wfile.closed: 704 self.wfile.flush() 705 self.wfile.close() 706 self.rfile.close() 707 708 709 class DatagramRequestHandler(BaseRequestHandler): 710 711 # XXX Regrettably, I cannot get this working on Linux; 712 # s.recvfrom() doesn't return a meaningful client address. 713 714 """Define self.rfile and self.wfile for datagram sockets.""" 715 716 def setup(self): 717 try: 718 from cStringIO import StringIO 719 except ImportError: 720 from StringIO import StringIO 721 self.packet, self.socket = self.request 722 self.rfile = StringIO(self.packet) 723 self.wfile = StringIO() 724 725 def finish(self): 726 self.socket.sendto(self.wfile.getvalue(), self.client_address)