diff options
Diffstat (limited to 'sleekxmpp/xmlstream')
| -rw-r--r-- | sleekxmpp/xmlstream/handler/base.py | 2 | ||||
| -rw-r--r-- | sleekxmpp/xmlstream/handler/callback.py | 4 | ||||
| -rw-r--r-- | sleekxmpp/xmlstream/scheduler.py | 87 | ||||
| -rw-r--r-- | sleekxmpp/xmlstream/stanzabase.py | 22 | ||||
| -rw-r--r-- | sleekxmpp/xmlstream/statemachine.py | 245 | ||||
| -rw-r--r-- | sleekxmpp/xmlstream/xmlstream.py | 323 |
6 files changed, 502 insertions, 181 deletions
diff --git a/sleekxmpp/xmlstream/handler/base.py b/sleekxmpp/xmlstream/handler/base.py index 5d55f4ee..a44edf0e 100644 --- a/sleekxmpp/xmlstream/handler/base.py +++ b/sleekxmpp/xmlstream/handler/base.py @@ -18,7 +18,7 @@ class BaseHandler(object): def match(self, xml): return self._matcher.match(xml) - def prerun(self, payload): + def prerun(self, payload): # what's the point of this if the payload is called again in run?? self._payload = payload def run(self, payload): diff --git a/sleekxmpp/xmlstream/handler/callback.py b/sleekxmpp/xmlstream/handler/callback.py index 49cfa14d..ea5acb5b 100644 --- a/sleekxmpp/xmlstream/handler/callback.py +++ b/sleekxmpp/xmlstream/handler/callback.py @@ -17,13 +17,15 @@ class Callback(base.BaseHandler): self._once = once self._instream = instream - def prerun(self, payload): + def prerun(self, payload): # prerun actually calls run?!? WTF! Then it gets run AGAIN! base.BaseHandler.prerun(self, payload) if self._instream: + logging.debug('callback "%s" prerun', self.name) self.run(payload, True) def run(self, payload, instream=False): if not self._instream or instream: + logging.debug('callback "%s" run', self.name) base.BaseHandler.run(self, payload) #if self._thread: # x = threading.Thread(name="Callback_%s" % self.name, target=self._pointer, args=(payload,)) diff --git a/sleekxmpp/xmlstream/scheduler.py b/sleekxmpp/xmlstream/scheduler.py new file mode 100644 index 00000000..40aaf695 --- /dev/null +++ b/sleekxmpp/xmlstream/scheduler.py @@ -0,0 +1,87 @@ +try: + import queue +except ImportError: + import Queue as queue +import time +import threading +import logging + +class Task(object): + """Task object for the Scheduler class""" + def __init__(self, name, seconds, callback, args=None, kwargs=None, repeat=False, qpointer=None): + self.name = name + self.seconds = seconds + self.callback = callback + self.args = args or tuple() + self.kwargs = kwargs or {} + self.repeat = repeat + self.next = time.time() + self.seconds + self.qpointer = qpointer + + def run(self): + if self.qpointer is not None: + self.qpointer.put(('schedule', self.callback, self.args)) + else: + self.callback(*self.args, **self.kwargs) + self.reset() + return self.repeat + + def reset(self): + self.next = time.time() + self.seconds + +class Scheduler(object): + """Threaded scheduler that allows for updates mid-execution unlike http://docs.python.org/library/sched.html#module-sched""" + def __init__(self, parentqueue=None): + self.addq = queue.Queue() + self.schedule = [] + self.thread = None + self.run = False + self.parentqueue = parentqueue + + def process(self, threaded=True): + if threaded: + self.thread = threading.Thread(name='shedulerprocess', target=self._process) + self.thread.start() + else: + self._process() + + def _process(self): + self.run = True + while self.run: + try: + wait = 1 + updated = False + if self.schedule: + wait = self.schedule[0].next - time.time() + try: + if wait <= 0.0: + newtask = self.addq.get(False) + else: + newtask = self.addq.get(True, wait) + except queue.Empty: + cleanup = [] + for task in self.schedule: + if time.time() >= task.next: + updated = True + if not task.run(): + cleanup.append(task) + else: + break + for task in cleanup: + x = self.schedule.pop(self.schedule.index(task)) + else: + updated = True + self.schedule.append(newtask) + finally: + if updated: self.schedule = sorted(self.schedule, key=lambda task: task.next) + except KeyboardInterrupt: + self.run = False + logging.debug("Quitting Scheduler thread") + if self.parentqueue is not None: + self.parentqueue.put(('quit', None, None)) + + def add(self, name, seconds, callback, args=None, kwargs=None, repeat=False, qpointer=None): + self.addq.put(Task(name, seconds, callback, args, kwargs, repeat, qpointer)) + + def quit(self): + self.run = False diff --git a/sleekxmpp/xmlstream/stanzabase.py b/sleekxmpp/xmlstream/stanzabase.py index 3f3f5e08..34513807 100644 --- a/sleekxmpp/xmlstream/stanzabase.py +++ b/sleekxmpp/xmlstream/stanzabase.py @@ -1,9 +1,9 @@ """ - SleekXMPP: The Sleek XMPP Library - Copyright (C) 2010 Nathanael C. Fritz - This file is part of SleekXMPP. + SleekXMPP: The Sleek XMPP Library + Copyright (C) 2010 Nathanael C. Fritz + This file is part of SleekXMPP. - See the file license.txt for copying permission. + See the file license.txt for copying permission. """ from xml.etree import cElementTree as ET import logging @@ -78,6 +78,9 @@ class ElementBase(tostring.ToString): def __iter__(self): self.idx = 0 return self + + def __bool__(self): + return True def __next__(self): self.idx += 1 @@ -319,6 +322,8 @@ class StanzaBase(ElementBase): def __init__(self, stream=None, xml=None, stype=None, sto=None, sfrom=None, sid=None): self.stream = stream + if stream is not None: + self.namespace = stream.default_ns ElementBase.__init__(self, xml) if stype is not None: self['type'] = stype @@ -326,8 +331,6 @@ class StanzaBase(ElementBase): self['to'] = sto if sfrom is not None: self['from'] = sfrom - if stream is not None: - self.namespace = stream.default_ns self.tag = "{%s}%s" % (self.namespace, self.name) def setType(self, value): @@ -380,6 +383,7 @@ class StanzaBase(ElementBase): def exception(self, e): logging.error(traceback.format_tb(e)) - def send(self): - self.stream.sendRaw(self.__str__()) - + def send(self, priority=False): + if priority: self.stream.sendPriorityRaw(self.__str__()) + else: self.stream.sendRaw(self.__str__()) + diff --git a/sleekxmpp/xmlstream/statemachine.py b/sleekxmpp/xmlstream/statemachine.py index fb7d1508..67b514a2 100644 --- a/sleekxmpp/xmlstream/statemachine.py +++ b/sleekxmpp/xmlstream/statemachine.py @@ -7,53 +7,228 @@ """ from __future__ import with_statement import threading +import time +import logging + class StateMachine(object): - def __init__(self, states=[], groups=[]): - self.lock = threading.Lock() - self.__state = {} - self.__default_state = {} - self.__group = {} + def __init__(self, states=[]): + self.lock = threading.Condition(threading.RLock()) + self.__states= [] self.addStates(states) - self.addGroups(groups) + self.__default_state = self.__states[0] + self.__current_state = self.__default_state def addStates(self, states): with self.lock: for state in states: - if state in self.__state or state in self.__group: - raise IndexError("The state or group '%s' is already in the StateMachine." % state) - self.__state[state] = states[state] - self.__default_state[state] = states[state] + if state in self.__states: + raise IndexError("The state '%s' is already in the StateMachine." % state) + self.__states.append( state ) - def addGroups(self, groups): - with self.lock: - for gstate in groups: - if gstate in self.__state or gstate in self.__group: - raise IndexError("The key or group '%s' is already in the StateMachine." % gstate) - for state in groups[gstate]: - if state in self.__state: - raise IndexError("The group %s contains a key %s which is not set in the StateMachine." % (gstate, state)) - self.__group[gstate] = groups[gstate] - - def set(self, state, status): + + def transition(self, from_state, to_state, wait=0.0, func=None, args=[], kwargs={} ): + ''' + Transition from the given `from_state` to the given `to_state`. + This method will return `True` if the state machine is now in `to_state`. It + will return `False` if a timeout occurred the transition did not occur. + If `wait` is 0 (the default,) this method returns immediately if the state machine + is not in `from_state`. + + If you want the thread to block and transition once the state machine to enters + `from_state`, set `wait` to a non-negative value. Note there is no 'block + indefinitely' flag since this leads to deadlock. If you want to wait indefinitely, + choose a reasonable value for `wait` (e.g. 20 seconds) and do so in a while loop like so: + + :: + + while not thread_should_exit and not state_machine.transition('disconnected', 'connecting', wait=20 ): + pass # timeout will occur every 20s unless transition occurs + if thread_should_exit: return + # perform actions here after successful transition + + This allows the thread to be responsive by setting `thread_should_exit=True`. + + The optional `func` argument allows the user to pass a callable operation which occurs + within the context of the state transition (e.g. while the state machine is locked.) + If `func` returns a True value, the transition will occur. If `func` returns a non- + True value or if an exception is thrown, the transition will not occur. Any thrown + exception is not caught by the state machine and is the caller's responsibility to handle. + If `func` completes normally, this method will return the value returned by `func.` If + values for `args` and `kwargs` are provided, they are expanded and passed like so: + `func( *args, **kwargs )`. + ''' + + return self.transition_any( (from_state,), to_state, wait=wait, + func=func, args=args, kwargs=kwargs ) + + + def transition_any(self, from_states, to_state, wait=0.0, func=None, args=[], kwargs={} ): + ''' + Transition from any of the given `from_states` to the given `to_state`. + ''' + + if not (isinstance(from_states,tuple) or isinstance(from_states,list)): + raise ValueError( "from_states should be a list or tuple" ) + + for state in from_states: + if not state in self.__states: + raise ValueError( "StateMachine does not contain from_state %s." % state ) + if not to_state in self.__states: + raise ValueError( "StateMachine does not contain to_state %s." % to_state ) + with self.lock: - if state in self.__state: - self.__state[state] = bool(status) + start = time.time() + while not self.__current_state in from_states: + # detect timeout: + if time.time() >= start + wait: return False + self.lock.wait(wait) + + if self.__current_state in from_states: # should always be True due to lock + + return_val = True + # Note that func might throw an exception, but that's OK, it aborts the transition + if func is not None: return_val = func(*args,**kwargs) + + # some 'false' value returned from func, + # indicating that transition should not occur: + if not return_val: return return_val + + logging.debug(' ==== TRANSITION %s -> %s', self.__current_state, to_state) + self.__current_state = to_state + self.lock.notify_all() + return return_val # some 'true' value returned by func or True if func was None else: - raise KeyError("StateMachine does not contain state %s." % state) - - def __getitem__(self, key): - if key in self.__group: - for state in self.__group[key]: - if not self.__state[state]: - return False - return True - return self.__state[key] + logging.error( "StateMachine bug!! The lock should ensure this doesn't happen!" ) + return False + + + def transition_ctx(self, from_state, to_state, wait=0.0): + ''' + Use the state machine as a context manager. The transition occurs on /exit/ from + the `with` context, so long as no exception is thrown. For example: + + :: + + with state_machine.transition_ctx('one','two', wait=5) as locked: + if locked: + # the state machine is currently locked in state 'one', and will + # transition to 'two' when the 'with' statement ends, so long as + # no exception is thrown. + print 'Currently locked in state one: %s' % state_machine['one'] + + else: + # The 'wait' timed out, and no lock has been acquired + print 'Timed out before entering state "one"' + + print 'Since no exception was thrown, we are now in state "two": %s' % state_machine['two'] + + + The other main difference between this method and `transition()` is that the + state machine is locked for the duration of the `with` statement. Normally, + after a `transition()` occurs, the state machine is immediately unlocked and + available to another thread to call `transition()` again. + ''' + + if not from_state in self.__states: + raise ValueError( "StateMachine does not contain from_state %s." % from_state ) + if not to_state in self.__states: + raise ValueError( "StateMachine does not contain to_state %s." % to_state ) + + return _StateCtx(self, from_state, to_state, wait) + - def __getattr__(self, attr): - return self.__getitem__(attr) + def ensure(self, state, wait=0.0): + ''' + Ensure the state machine is currently in `state`, or wait until it enters `state`. + ''' + return self.ensure_any( (state,), wait=wait ) + + + def ensure_any(self, states, wait=0.0): + ''' + Ensure we are currently in one of the given `states` + ''' + if not (isinstance(states,tuple) or isinstance(states,list)): + raise ValueError('states arg should be a tuple or list') + + for state in states: + if not state in self.__states: + raise ValueError( "StateMachine does not contain state '%s'" % state ) + + with self.lock: + start = time.time() + while not self.__current_state in states: + # detect timeout: + if time.time() >= start + wait: return False + self.lock.wait(wait) + return self.__current_state in states # should always be True due to lock + def reset(self): - self.__state = self.__default_state + # TODO need to lock before calling this? + self.transition(self.__current_state, self._default_state) + + + def _set_state(self, state): #unsynchronized, only call internally after lock is acquired + self.__current_state = state + return state + + + def current_state(self): + ''' + Return the current state name. + ''' + return self.__current_state + + + def __getitem__(self, state): + ''' + Non-blocking, non-synchronized test to determine if we are in the given state. + Use `StateMachine.ensure(state)` to wait until the machine enters a certain state. + ''' + return self.__current_state == state + + def __str__(self): + return "".join(( "StateMachine(", ','.join(self.__states), "): ", self.__current_state )) + + + +class _StateCtx: + + def __init__( self, state_machine, from_state, to_state, wait ): + self.state_machine = state_machine + self.from_state = from_state + self.to_state = to_state + self.wait = wait + self._timeout = False + + def __enter__(self): + self.state_machine.lock.acquire() + start = time.time() + while not self.state_machine[ self.from_state ]: + # detect timeout: + if time.time() >= start + self.wait: + logging.debug('StateMachine timeout while waiting for state: %s', self.from_state ) + self._timeout = True # to indicate we should not transition + return False + self.state_machine.lock.wait(self.wait) + + logging.debug('StateMachine entered context in state: %s', + self.state_machine.current_state() ) + return True + + def __exit__(self, exc_type, exc_val, exc_tb): + if exc_val is not None: + logging.exception( "StateMachine exception in context, remaining in state: %s\n%s:%s", + self.state_machine.current_state(), exc_type.__name__, exc_val ) + elif not self._timeout: + logging.debug(' ==== TRANSITION %s -> %s', + self.state_machine.current_state(), self.to_state) + self.state_machine._set_state( self.to_state ) + + self.state_machine.lock.notify_all() + self.state_machine.lock.release() + return False # re-raise any exception diff --git a/sleekxmpp/xmlstream/xmlstream.py b/sleekxmpp/xmlstream/xmlstream.py index 025884b7..a8bcac00 100644 --- a/sleekxmpp/xmlstream/xmlstream.py +++ b/sleekxmpp/xmlstream/xmlstream.py @@ -1,9 +1,9 @@ """ - SleekXMPP: The Sleek XMPP Library - Copyright (C) 2010 Nathanael C. Fritz - This file is part of SleekXMPP. + SleekXMPP: The Sleek XMPP Library + Copyright (C) 2010 Nathanael C. Fritz + This file is part of SleekXMPP. - See the file license.txt for copying permission. + See the file license.txt for copying permission. """ from __future__ import with_statement, unicode_literals @@ -16,12 +16,14 @@ from . stanzabase import StanzaBase from xml.etree import cElementTree from xml.parsers import expat import logging +import random import socket import threading import time import traceback import types import xml.sax.saxutils +from . import scheduler HANDLER_THREADS = 1 @@ -45,6 +47,10 @@ class CloseStream(Exception): stanza_extensions = {} +RECONNECT_MAX_DELAY = 3600 +RECONNECT_QUIESCE_FACTOR = 1.6180339887498948 # Phi +RECONNECT_QUIESCE_JITTER = 0.11962656472 # molar Planck constant times c, joule meter/mole + class XMLStream(object): "A connection manager with XML events." @@ -52,8 +58,9 @@ class XMLStream(object): global ssl_support self.ssl_support = ssl_support self.escape_quotes = escape_quotes - self.state = statemachine.StateMachine() - self.state.addStates({'connected':False, 'is client':False, 'ssl':False, 'tls':False, 'reconnect':True, 'processing':False, 'disconnecting':False}) #set initial states + self.state = statemachine.StateMachine(('disconnected','connecting', + 'connected')) + self.should_reconnect = True self.setSocket(socket) self.address = (host, int(port)) @@ -69,12 +76,14 @@ class XMLStream(object): self.filesocket = None self.use_ssl = False self.use_tls = False + self.ca_certs=None self.stream_header = "<stream>" self.stream_footer = "</stream>" self.eventqueue = queue.Queue() - self.sendqueue = queue.Queue() + self.sendqueue = queue.PriorityQueue() + self.scheduler = scheduler.Scheduler(self.eventqueue) self.namespace_map = {} @@ -83,45 +92,77 @@ class XMLStream(object): def setSocket(self, socket): "Set the socket" self.socket = socket - if socket is not None: + if socket is not None and self.state.transition('disconnected','connecting'): self.filesocket = socket.makefile('rb', 0) # ElementTree.iterparse requires a file. 0 buffer files have to be binary - self.state.set('connected', True) - + self.state.transition('connecting','connected') def setFileSocket(self, filesocket): self.filesocket = filesocket - def connect(self, host='', port=0, use_ssl=False, use_tls=True): - "Link to connectTCP" - return self.connectTCP(host, port, use_ssl, use_tls) + def connect(self, host='', port=0, use_ssl=None, use_tls=None): + "Establish a socket connection to the given XMPP server." + + if not self.state.transition('disconnected','connected', + func=self.connectTCP, args=[host, port, use_ssl, use_tls] ): + + if self.state['connected']: logging.debug('Already connected') + else: logging.warning("Connection failed" ) + return False + + logging.debug('Connection complete.') + return True + + # TODO currently a caller can't distinguish between "connection failed" and + # "we're already trying to connect from another thread" def connectTCP(self, host='', port=0, use_ssl=None, use_tls=None, reattempt=True): "Connect and create socket" - while reattempt and not self.state['connected']: - if host and port: - self.address = (host, int(port)) - if use_ssl is not None: - self.use_ssl = use_ssl - if use_tls is not None: - self.use_tls = use_tls - self.state.set('is client', True) - if sys.version_info < (3, 0): - self.socket = filesocket.Socket26(socket.AF_INET, socket.SOCK_STREAM) - else: - self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) - self.socket.settimeout(None) - if self.use_ssl and self.ssl_support: - logging.debug("Socket Wrapped for SSL") - self.socket = ssl.wrap_socket(self.socket) + + # Note that this is thread-safe by merit of being called solely from connect() which + # holds the state lock. + + delay = 1.0 # reconnection delay + while self.run: + logging.debug('connecting....') try: + if host and port: + self.address = (host, int(port)) + if use_ssl is not None: + self.use_ssl = use_ssl + if use_tls is not None: + # TODO this variable doesn't seem to be used for anything! + self.use_tls = use_tls + if sys.version_info < (3, 0): + self.socket = filesocket.Socket26(socket.AF_INET, socket.SOCK_STREAM) + else: + self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) + self.socket.settimeout(None) #10) + + if self.use_ssl and self.ssl_support: + logging.debug("Socket Wrapped for SSL") + self.socket = ssl.wrap_socket(self.socket,ca_certs=self.ca_certs) + self.socket.connect(self.address) - #self.filesocket = self.socket.makefile('rb', 0) self.filesocket = self.socket.makefile('rb', 0) - self.state.set('connected', True) + return True + except socket.error as serr: - logging.error("Could not connect. Socket Error #%s: %s" % (serr.errno, serr.strerror)) - time.sleep(1) + logging.exception("Socket Error #%s: %s", serr.errno, serr.strerror) + if not reattempt: return False + except: + logging.exception("Connection error") + if not reattempt: return False + + # quiesce if rconnection fails: + # This algorithm based loosely on Twisted internet.protocol + # http://twistedmatrix.com/trac/browser/trunk/twisted/internet/protocol.py#L310 + delay = min(delay * RECONNECT_QUIESCE_FACTOR, RECONNECT_MAX_DELAY) + delay = random.normalvariate(delay, delay * RECONNECT_QUIESCE_JITTER) + logging.debug('Waiting %fs until next reconnect attempt...', delay) + time.sleep(delay) + + def connectUnix(self, filepath): "Connect to Unix file and create socket" @@ -130,14 +171,19 @@ class XMLStream(object): "Handshakes for TLS" if self.ssl_support: logging.info("Negotiating TLS") - self.realsocket = self.socket - self.socket = ssl.wrap_socket(self.socket, ssl_version=ssl.PROTOCOL_TLSv1, do_handshake_on_connect=False) +# self.realsocket = self.socket # NOT USED + self.socket = ssl.wrap_socket(self.socket, + ssl_version=ssl.PROTOCOL_TLSv1, + do_handshake_on_connect=False, + ca_certs=self.ca_certs) self.socket.do_handshake() if sys.version_info < (3,0): from . filesocket import filesocket self.filesocket = filesocket(self.socket) else: self.filesocket = self.socket.makefile('rb', 0) + + logging.debug("TLS negotitation successful") return True else: logging.warning("Tried to enable TLS, but ssl module not found.") @@ -145,67 +191,56 @@ class XMLStream(object): raise RestartStream() def process(self, threaded=True): + self.scheduler.process(threaded=True) + self.run = True for t in range(0, HANDLER_THREADS): - self.__thread['eventhandle%s' % t] = threading.Thread(name='eventhandle%s' % t, target=self._eventRunner) - self.__thread['eventhandle%s' % t].start() - self.__thread['sendthread'] = threading.Thread(name='sendthread', target=self._sendThread) - self.__thread['sendthread'].start() + th = threading.Thread(name='eventhandle%s' % t, target=self._eventRunner) + th.setDaemon(True) + self.__thread['eventhandle%s' % t] = th + th.start() + th = threading.Thread(name='sendthread', target=self._sendThread) + th.setDaemon(True) + self.__thread['sendthread'] = th + th.start() if threaded: - self.__thread['process'] = threading.Thread(name='process', target=self._process) - self.__thread['process'].start() + th = threading.Thread(name='process', target=self._process) + th.setDaemon(True) + self.__thread['process'] = th + th.start() else: self._process() - def schedule(self, seconds, handler, args=None): - threading.Timer(seconds, handler, args).start() + def schedule(self, name, seconds, callback, args=None, kwargs=None, repeat=False): + self.scheduler.add(name, seconds, callback, args, kwargs, repeat, qpointer=self.eventqueue) def _process(self): "Start processing the socket." - firstrun = True - while self.run and (firstrun or self.state['reconnect']): - self.state.set('processing', True) - firstrun = False + logging.debug('Process thread starting...') + while self.run: + if not self.state.ensure('connected',wait=2): continue try: - if self.state['is client']: - self.sendRaw(self.stream_header) - while self.run and self.__readXML(): - if self.state['is client']: - self.sendRaw(self.stream_header) - except KeyboardInterrupt: - logging.debug("Keyboard Escape Detected") - self.state.set('processing', False) - self.state.set('reconnect', False) - self.disconnect() - self.run = False - self.eventqueue.put(('quit', None, None)) - return + self.sendPriorityRaw(self.stream_header) + while self.run and self.__readXML(): pass + except socket.timeout: + logging.debug('socket rcv timeout') + pass except CloseStream: - return - except SystemExit: + # TODO warn that the listener thread is exiting!!! + pass + except RestartStream: + logging.debug("Restarting stream...") + continue # DON'T re-initialize the stream -- this exception is sent + # specifically when we've initialized TLS and need to re-send the <stream> header. + except (KeyboardInterrupt, SystemExit): + logging.debug("System interrupt detected") + self.shutdown() self.eventqueue.put(('quit', None, None)) - return - except socket.error: - if not self.state.reconnect: - return - else: - self.state.set('processing', False) - traceback.print_exc() - self.disconnect(reconnect=True) except: - if not self.state.reconnect: - return - else: - self.state.set('processing', False) - traceback.print_exc() + logging.exception('Unexpected error in RCV thread') + if self.should_reconnect: self.disconnect(reconnect=True) - if self.state['reconnect']: - self.reconnect() - self.state.set('processing', False) - self.eventqueue.put(('quit', None, None)) - #self.__thread['readXML'] = threading.Thread(name='readXML', target=self.__readXML) - #self.__thread['readXML'].start() - #self.__thread['spawnEvents'] = threading.Thread(name='spawnEvents', target=self.__spawnEvents) - #self.__thread['spawnEvents'].start() + + logging.debug('Quitting Process thread') def __readXML(self): "Parses the incoming stream, adding to xmlin queue as it goes" @@ -218,82 +253,94 @@ class XMLStream(object): if edepth == 0: # and xmlobj.tag.split('}', 1)[-1] == self.basetag: if event == b'start': root = xmlobj + logging.debug('handling start stream') self.start_stream_handler(root) if event == b'end': edepth += -1 if edepth == 0 and event == b'end': - self.disconnect(reconnect=self.state['reconnect']) + # what is this case exactly? Premature EOF? + logging.debug("Ending readXML loop") return False elif edepth == 1: #self.xmlin.put(xmlobj) - try: - self.__spawnEvent(xmlobj) - except RestartStream: - return True - except CloseStream: - return False - if root: - root.clear() + self.__spawnEvent(xmlobj) + if root: root.clear() if event == b'start': edepth += 1 + logging.debug("Exiting readXML loop") + return False def _sendThread(self): + logging.debug('send thread starting...') while self.run: - data = self.sendqueue.get(True) - logging.debug("SEND: %s" % data) + if not self.state.ensure('connected',wait=2): continue + + data = None try: - self.socket.send(data.encode('utf-8')) - #self.socket.send(bytes(data, "utf-8")) - #except socket.error,(errno, strerror): + data = self.sendqueue.get(True,5)[1] + logging.debug("SEND: %s" % data) + self.socket.sendall(data.encode('utf-8')) + except queue.Empty: +# logging.debug('Nothing on send queue') + pass + except socket.timeout: + # this is to prevent a thread blocked indefinitely + logging.debug('timeout sending packet data') except: logging.warning("Failed to send %s" % data) - self.state.set('connected', False) - if self.state.reconnect: - logging.error("Disconnected. Socket Error.") - traceback.print_exc() + logging.exception("Socket error in SEND thread") + # TODO it's somewhat unsafe for the sender thread to assume it can just + # re-intitialize the connection, since the receiver thread could be doing + # the same thing concurrently. Oops! The safer option would be to throw + # some sort of event that could be handled by a common thread or the reader + # thread to perform reconnect and then re-initialize the handler threads as well. + if self.should_reconnect: self.disconnect(reconnect=True) def sendRaw(self, data): - self.sendqueue.put(data) + self.sendqueue.put((1, data)) + return True + + def sendPriorityRaw(self, data): + self.sendqueue.put((0, data)) return True def disconnect(self, reconnect=False): - self.state.set('reconnect', reconnect) - if self.state['disconnecting']: + if not self.state.transition('connected','disconnected'): + logging.warning("Already disconnected.") return - if not self.state['reconnect']: - logging.debug("Disconnecting...") - self.state.set('disconnecting', True) - self.run = False - if self.state['connected']: - self.sendRaw(self.stream_footer) - time.sleep(1) - #send end of stream - #wait for end of stream back + logging.debug("Disconnecting...") + self.sendPriorityRaw(self.stream_footer) + time.sleep(5) + #send end of stream + #wait for end of stream back try: +# self.socket.shutdown(socket.SHUT_RDWR) self.socket.close() + except socket.error as (errno,strerror): + logging.exception("Error while disconnecting. Socket Error #%s: %s" % (errno, strerror)) + try: self.filesocket.close() - self.socket.shutdown(socket.SHUT_RDWR) - except socket.error as serr: - #logging.warning("Error while disconnecting. Socket Error #%s: %s" % (errno, strerror)) - #thread.exit_thread() - pass - if self.state['processing']: - #raise CloseStream - pass - - def reconnect(self): - self.state.set('tls',False) - self.state.set('ssl',False) - time.sleep(1) - self.connect() + except socket.error as (errno,strerror): + logging.exception("Error closing filesocket.") + + if reconnect: self.connect() + def shutdown(self): + ''' + Disconnects and shuts down all event threads. + ''' + self.disconnect() + self.run = False + self.scheduler.run = False + def incoming_filter(self, xmlobj): return xmlobj - + def __spawnEvent(self, xmlobj): "watching xmlOut and processes handlers" #convert XML into Stanza + # TODO surround this log statement with an if, it's expensive logging.debug("RECV: %s" % cElementTree.tostring(xmlobj)) xmlobj = self.incoming_filter(xmlobj) stanza = None @@ -305,48 +352,54 @@ class XMLStream(object): if stanza is None: stanza = StanzaBase(self, xmlobj) unhandled = True + # TODO inefficient linear search; performance might be improved by hashtable lookup for handler in self.__handlers: if handler.match(stanza): + logging.debug('matched stanza to handler %s', handler.name) handler.prerun(stanza) self.eventqueue.put(('stanza', handler, stanza)) - if handler.checkDelete(): self.__handlers.pop(self.__handlers.index(handler)) + if handler.checkDelete(): + logging.debug('deleting callback %s', handler.name) + self.__handlers.pop(self.__handlers.index(handler)) unhandled = False if unhandled: stanza.unhandled() #loop through handlers and test match #spawn threads as necessary, call handlers, sending Stanza - + def _eventRunner(self): logging.debug("Loading event runner") while self.run: try: event = self.eventqueue.get(True, timeout=5) except queue.Empty: +# logging.debug('Nothing on event queue') event = None if event is not None: etype = event[0] handler = event[1] args = event[2:] - #etype, handler, *args = event #python 3.x way + #etype, handler, *args = event #python 3.x way if etype == 'stanza': try: handler.run(args[0]) except Exception as e: - traceback.print_exc() + logging.exception("Exception in event handler") args[0].exception(e) elif etype == 'sched': try: + #handler(*args[0]) handler.run(*args) except: logging.error(traceback.format_exc()) elif etype == 'quit': logging.debug("Quitting eventRunner thread") return False - + def registerHandler(self, handler, before=None, after=None): "Add handler with matcher class and parameters." self.__handlers.append(handler) - + def removeHandler(self, name): "Removes the handler." idx = 0 @@ -432,4 +485,4 @@ class XMLStream(object): def start_stream_handler(self, xml): """Meant to be overridden""" - pass + logging.warn("No start stream handler has been implemented.") |