""" Core logic (uri, daemon, proxy stuff). Pyro - Python Remote Objects. Copyright by Irmen de Jong (irmen@razorvine.net). """ from __future__ import print_function, division import inspect import re import logging import sys import ssl import os import time import threading import uuid import base64 import warnings import socket import random from Pyro4 import errors, socketutil, util, constants, message, futures from Pyro4.configuration import config __all__ = ["URI", "Proxy", "Daemon", "current_context", "callback", "batch", "asyncproxy", "expose", "behavior", "oneway", "SerializedBlob", "_resolve", "_locateNS"] if sys.version_info >= (3, 0): basestring = str log = logging.getLogger("Pyro4.core") class URI(object): """ Pyro object URI (universal resource identifier). The uri format is like this: ``PYRO:objectid@location`` where location is one of: - ``hostname:port`` (tcp/ip socket on given port) - ``./u:sockname`` (Unix domain socket on localhost) There is also a 'Magic format' for simple name resolution using Name server: ``PYRONAME:objectname[@location]`` (optional name server location, can also omit location port) And one that looks up things in the name server by metadata: ``PYROMETA:meta1,meta2,...[@location]`` (optional name server location, can also omit location port) You can write the protocol in lowercase if you like (``pyro:...``) but it will automatically be converted to uppercase internally. """ uriRegEx = re.compile(r"(?P[Pp][Yy][Rr][Oo][a-zA-Z]*):(?P\S+?)(@(?P.+))?$") def __init__(self, uri): if isinstance(uri, URI): state = uri.__getstate__() self.__setstate__(state) return if not isinstance(uri, basestring): raise TypeError("uri parameter object is of wrong type") self.sockname = self.host = self.port = None match = self.uriRegEx.match(uri) if not match: raise errors.PyroError("invalid uri") self.protocol = match.group("protocol").upper() self.object = match.group("object") location = match.group("location") if self.protocol == "PYRONAME": self._parseLocation(location, config.NS_PORT) elif self.protocol == "PYRO": if not location: raise errors.PyroError("invalid uri") self._parseLocation(location, None) elif self.protocol == "PYROMETA": self.object = set(m.strip() for m in self.object.split(",")) self._parseLocation(location, config.NS_PORT) else: raise errors.PyroError("invalid uri (protocol)") def _parseLocation(self, location, defaultPort): if not location: return if location.startswith("./u:"): self.sockname = location[4:] if (not self.sockname) or ':' in self.sockname: raise errors.PyroError("invalid uri (location)") else: if location.startswith("["): # ipv6 if location.startswith("[["): # possible mistake: double-bracketing raise errors.PyroError("invalid ipv6 address: enclosed in too many brackets") ipv6locationmatch = re.match(r"\[([0-9a-fA-F:%]+)](:(\d+))?", location) if not ipv6locationmatch: raise errors.PyroError("invalid ipv6 address: the part between brackets must be a numeric ipv6 address") self.host, _, self.port = ipv6locationmatch.groups() else: self.host, _, self.port = location.partition(":") if not self.port: self.port = defaultPort try: self.port = int(self.port) except (ValueError, TypeError): raise errors.PyroError("invalid port in uri, port=" + str(self.port)) @staticmethod def isUnixsockLocation(location): """determine if a location string is for a Unix domain socket""" return location.startswith("./u:") @property def location(self): """property containing the location string, for instance ``"servername.you.com:5555"``""" if self.host: if ":" in self.host: # ipv6 return "[%s]:%d" % (self.host, self.port) else: return "%s:%d" % (self.host, self.port) elif self.sockname: return "./u:" + self.sockname else: return None def asString(self): """the string representation of this object""" if self.protocol == "PYROMETA": result = "PYROMETA:" + ",".join(self.object) else: result = self.protocol + ":" + self.object location = self.location if location: result += "@" + location return result def __str__(self): string = self.asString() if sys.version_info < (3, 0) and type(string) is unicode: return string.encode("ascii", "replace") return string def __unicode__(self): return self.asString() def __repr__(self): return "<%s.%s at 0x%x; %s>" % (self.__class__.__module__, self.__class__.__name__, id(self), str(self)) def __eq__(self, other): if not isinstance(other, URI): return False return (self.protocol, self.object, self.sockname, self.host, self.port) ==\ (other.protocol, other.object, other.sockname, other.host, other.port) def __ne__(self, other): return not self.__eq__(other) def __hash__(self): return hash((self.protocol, str(self.object), self.sockname, self.host, self.port)) # note: getstate/setstate are not needed if we use pickle protocol 2, # but this way it helps pickle to make the representation smaller by omitting all attribute names. def __getstate__(self): return self.protocol, self.object, self.sockname, self.host, self.port def __setstate__(self, state): self.protocol, self.object, self.sockname, self.host, self.port = state def __getstate_for_dict__(self): return self.__getstate__() def __setstate_from_dict__(self, state): self.__setstate__(state) class _RemoteMethod(object): """method call abstraction""" def __init__(self, send, name, max_retries): self.__send = send self.__name = name self.__max_retries = max_retries def __getattr__(self, name): return _RemoteMethod(self.__send, "%s.%s" % (self.__name, name), self.__max_retries) def __call__(self, *args, **kwargs): for attempt in range(self.__max_retries + 1): try: return self.__send(self.__name, args, kwargs) except (errors.ConnectionClosedError, errors.TimeoutError): # only retry for recoverable network errors if attempt >= self.__max_retries: # last attempt, raise the exception raise class Proxy(object): """ Pyro proxy for a remote object. Intercepts method calls and dispatches them to the remote object. .. automethod:: _pyroBind .. automethod:: _pyroRelease .. automethod:: _pyroReconnect .. automethod:: _pyroBatch .. automethod:: _pyroAsync .. automethod:: _pyroAnnotations .. automethod:: _pyroResponseAnnotations .. automethod:: _pyroValidateHandshake .. autoattribute:: _pyroTimeout .. autoattribute:: _pyroHmacKey .. attribute:: _pyroMaxRetries Number of retries to perform on communication calls by this proxy, allows you to override the default setting. .. attribute:: _pyroSerializer Name of the serializer to use by this proxy, allows you to override the default setting. .. attribute:: _pyroHandshake The data object that should be sent in the initial connection handshake message. Can be any serializable object. """ __pyroAttributes = frozenset( ["__getnewargs__", "__getnewargs_ex__", "__getinitargs__", "_pyroConnection", "_pyroUri", "_pyroOneway", "_pyroMethods", "_pyroAttrs", "_pyroTimeout", "_pyroSeq", "_pyroHmacKey", "_pyroRawWireResponse", "_pyroHandshake", "_pyroMaxRetries", "_pyroSerializer", "_Proxy__async", "_Proxy__pyroHmacKey", "_Proxy__pyroTimeout", "_Proxy__pyroConnLock"]) def __init__(self, uri, connected_socket=None): if connected_socket: uri = URI("PYRO:" + uri + "@<>:0") if isinstance(uri, basestring): uri = URI(uri) elif not isinstance(uri, URI): raise TypeError("expected Pyro URI") self._pyroUri = uri self._pyroConnection = None self._pyroSerializer = None # can be set to the name of a serializer to override the global one per-proxy self._pyroMethods = set() # all methods of the remote object, gotten from meta-data self._pyroAttrs = set() # attributes of the remote object, gotten from meta-data self._pyroOneway = set() # oneway-methods of the remote object, gotten from meta-data self._pyroSeq = 0 # message sequence number self._pyroRawWireResponse = False # internal switch to enable wire level responses self._pyroHandshake = "hello" # the data object that should be sent in the initial connection handshake message self._pyroMaxRetries = config.MAX_RETRIES self.__pyroHmacKey = None self.__pyroTimeout = config.COMMTIMEOUT self.__pyroConnLock = threading.RLock() util.get_serializer(config.SERIALIZER) # assert that the configured serializer is available self.__async = False current_context.annotations = {} current_context.response_annotations = {} if connected_socket: self.__pyroCreateConnection(False, connected_socket) @property def _pyroHmacKey(self): """the HMAC key (bytes) that this proxy uses""" return self.__pyroHmacKey @_pyroHmacKey.setter def _pyroHmacKey(self, value): # if needed, convert the hmac value to bytes first if value and sys.version_info >= (3, 0) and type(value) is not bytes: value = value.encode("utf-8") # convert to bytes self.__pyroHmacKey = value def __del__(self): if hasattr(self, "_pyroConnection"): self._pyroRelease() def __getattr__(self, name): if name in Proxy.__pyroAttributes: # allows it to be safely pickled raise AttributeError(name) if config.METADATA: # get metadata if it's not there yet if not self._pyroMethods and not self._pyroAttrs: self._pyroGetMetadata() if name in self._pyroAttrs: return self._pyroInvoke("__getattr__", (name,), None) if config.METADATA and name not in self._pyroMethods: # client side check if the requested attr actually exists raise AttributeError("remote object '%s' has no exposed attribute or method '%s'" % (self._pyroUri, name)) if self.__async: return _AsyncRemoteMethod(self, name, self._pyroMaxRetries) return _RemoteMethod(self._pyroInvoke, name, self._pyroMaxRetries) def __setattr__(self, name, value): if name in Proxy.__pyroAttributes: return super(Proxy, self).__setattr__(name, value) # one of the special pyro attributes if config.METADATA: # get metadata if it's not there yet if not self._pyroMethods and not self._pyroAttrs: self._pyroGetMetadata() if name in self._pyroAttrs: return self._pyroInvoke("__setattr__", (name, value), None) # remote attribute if config.METADATA: # client side validation if the requested attr actually exists raise AttributeError("remote object '%s' has no exposed attribute '%s'" % (self._pyroUri, name)) # metadata disabled, just treat it as a local attribute on the proxy: return super(Proxy, self).__setattr__(name, value) def __repr__(self): if self._pyroConnection: connected = "connected " + self._pyroConnection.family() else: connected = "not connected" return "<%s.%s at 0x%x; %s; for %s>" % (self.__class__.__module__, self.__class__.__name__, id(self), connected, self._pyroUri) def __unicode__(self): return str(self) def __getstate_for_dict__(self): encodedHmac = None if self._pyroHmacKey is not None: encodedHmac = "b64:" + (base64.b64encode(self._pyroHmacKey).decode("ascii")) # for backwards compatibility reasons we also put the timeout and maxretries into the state return self._pyroUri.asString(), tuple(self._pyroOneway), tuple(self._pyroMethods), tuple(self._pyroAttrs),\ self.__pyroTimeout, encodedHmac, self._pyroHandshake, self._pyroMaxRetries, self._pyroSerializer def __setstate_from_dict__(self, state): uri = URI(state[0]) oneway = set(state[1]) methods = set(state[2]) attrs = set(state[3]) timeout = state[4] hmac_key = state[5] handshake = state[6] max_retries = state[7] serializer = None if len(state) < 9 else state[8] if hmac_key: if hmac_key.startswith("b64:"): hmac_key = base64.b64decode(hmac_key[4:].encode("ascii")) else: raise errors.ProtocolError("hmac encoding error") self.__setstate__((uri, oneway, methods, attrs, timeout, hmac_key, handshake, max_retries, serializer)) def __getstate__(self): # for backwards compatibility reasons we also put the timeout and maxretries into the state return self._pyroUri, self._pyroOneway, self._pyroMethods, self._pyroAttrs, self.__pyroTimeout, \ self._pyroHmacKey, self._pyroHandshake, self._pyroMaxRetries, self._pyroSerializer def __setstate__(self, state): # Note that the timeout and maxretries are also part of the state (for backwards compatibility reasons), # but we're not using them here. Instead we get the configured values from the 'local' config. self._pyroUri, self._pyroOneway, self._pyroMethods, self._pyroAttrs, _, self._pyroHmacKey, self._pyroHandshake = state[:7] self._pyroSerializer = None if len(state) < 9 else state[8] self.__pyroTimeout = config.COMMTIMEOUT self._pyroMaxRetries = config.MAX_RETRIES self._pyroConnection = None self._pyroSeq = 0 self._pyroRawWireResponse = False self.__pyroConnLock = threading.RLock() self.__async = False def __copy__(self): uriCopy = URI(self._pyroUri) p = type(self)(uriCopy) p._pyroOneway = set(self._pyroOneway) p._pyroMethods = set(self._pyroMethods) p._pyroAttrs = set(self._pyroAttrs) p._pyroSerializer = self._pyroSerializer p._pyroTimeout = self._pyroTimeout p._pyroHandshake = self._pyroHandshake p._pyroHmacKey = self._pyroHmacKey p._pyroRawWireResponse = self._pyroRawWireResponse p._pyroMaxRetries = self._pyroMaxRetries p.__async = self.__async return p def __enter__(self): return self def __exit__(self, exc_type, exc_value, traceback): self._pyroRelease() def __eq__(self, other): if other is self: return True return isinstance(other, Proxy) and other._pyroUri == self._pyroUri def __ne__(self, other): if other and isinstance(other, Proxy): return other._pyroUri != self._pyroUri return True def __hash__(self): return hash(self._pyroUri) def __dir__(self): result = dir(self.__class__) + list(self.__dict__.keys()) return sorted(set(result) | self._pyroMethods | self._pyroAttrs) def _pyroRelease(self): """release the connection to the pyro daemon""" with self.__pyroConnLock: if self._pyroConnection is not None: if self._pyroConnection.keep_open: return self._pyroConnection.close() self._pyroConnection = None log.debug("connection released") def _pyroBind(self): """ Bind this proxy to the exact object from the uri. That means that the proxy's uri will be updated with a direct PYRO uri, if it isn't one yet. If the proxy is already bound, it will not bind again. """ return self.__pyroCreateConnection(True) def __pyroGetTimeout(self): return self.__pyroTimeout def __pyroSetTimeout(self, timeout): self.__pyroTimeout = timeout if self._pyroConnection is not None: self._pyroConnection.timeout = timeout _pyroTimeout = property(__pyroGetTimeout, __pyroSetTimeout, doc=""" The timeout in seconds for calls on this proxy. Defaults to ``None``. If the timeout expires before the remote method call returns, Pyro will raise a :exc:`Pyro4.errors.TimeoutError`""") def _pyroInvoke(self, methodname, vargs, kwargs, flags=0, objectId=None): """perform the remote method call communication""" current_context.response_annotations = {} with self.__pyroConnLock: if self._pyroConnection is None: self.__pyroCreateConnection() serializer = util.get_serializer(self._pyroSerializer or config.SERIALIZER) objectId = objectId or self._pyroConnection.objectId annotations = self.__annotations() if vargs and isinstance(vargs[0], SerializedBlob): # special serialization of a 'blob' that stays serialized data, compressed, flags = self.__serializeBlobArgs(vargs, kwargs, annotations, flags, objectId, methodname, serializer) else: # normal serialization of the remote call data, compressed = serializer.serializeCall(objectId, methodname, vargs, kwargs, compress=config.COMPRESSION) if compressed: flags |= message.FLAGS_COMPRESSED if methodname in self._pyroOneway: flags |= message.FLAGS_ONEWAY self._pyroSeq = (self._pyroSeq + 1) & 0xffff msg = message.Message(message.MSG_INVOKE, data, serializer.serializer_id, flags, self._pyroSeq, annotations=annotations, hmac_key=self._pyroHmacKey) if config.LOGWIRE: _log_wiredata(log, "proxy wiredata sending", msg) try: self._pyroConnection.send(msg.to_bytes()) del msg # invite GC to collect the object, don't wait for out-of-scope if flags & message.FLAGS_ONEWAY: return None # oneway call, no response data else: msg = message.Message.recv(self._pyroConnection, [message.MSG_RESULT], hmac_key=self._pyroHmacKey) if config.LOGWIRE: _log_wiredata(log, "proxy wiredata received", msg) self.__pyroCheckSequence(msg.seq) if msg.serializer_id != serializer.serializer_id: error = "invalid serializer in response: %d" % msg.serializer_id log.error(error) raise errors.SerializeError(error) if msg.annotations: current_context.response_annotations = msg.annotations self._pyroResponseAnnotations(msg.annotations, msg.type) if self._pyroRawWireResponse: msg.decompress_if_needed() return msg data = serializer.deserializeData(msg.data, compressed=msg.flags & message.FLAGS_COMPRESSED) if msg.flags & message.FLAGS_ITEMSTREAMRESULT: streamId = bytes(msg.annotations.get("STRM", b"")).decode() if not streamId: raise errors.ProtocolError("result of call is an iterator, but the server is not configured to allow streaming") return _StreamResultIterator(streamId, self) if msg.flags & message.FLAGS_EXCEPTION: if sys.platform == "cli": util.fixIronPythonExceptionForPickle(data, False) raise data # if you see this in your traceback, you should probably inspect the remote traceback as well else: return data except (errors.CommunicationError, KeyboardInterrupt): # Communication error during read. To avoid corrupt transfers, we close the connection. # Otherwise we might receive the previous reply as a result of a new method call! # Special case for keyboardinterrupt: people pressing ^C to abort the client # may be catching the keyboardinterrupt in their code. We should probably be on the # safe side and release the proxy connection in this case too, because they might # be reusing the proxy object after catching the exception... self._pyroRelease() raise def __pyroCheckSequence(self, seq): if seq != self._pyroSeq: err = "invoke: reply sequence out of sync, got %d expected %d" % (seq, self._pyroSeq) log.error(err) raise errors.ProtocolError(err) def __pyroCreateConnection(self, replaceUri=False, connected_socket=None): """ Connects this proxy to the remote Pyro daemon. Does connection handshake. Returns true if a new connection was made, false if an existing one was already present. """ def connect_and_handshake(conn): try: if self._pyroConnection is not None: return False # already connected if config.SSL: sslContext = socketutil.getSSLcontext(clientcert=config.SSL_CLIENTCERT, clientkey=config.SSL_CLIENTKEY, keypassword=config.SSL_CLIENTKEYPASSWD, cacerts=config.SSL_CACERTS) else: sslContext = None sock = socketutil.createSocket(connect=connect_location, reuseaddr=config.SOCK_REUSE, timeout=self.__pyroTimeout, nodelay=config.SOCK_NODELAY, sslContext=sslContext) conn = socketutil.SocketConnection(sock, uri.object) # Do handshake. serializer = util.get_serializer(self._pyroSerializer or config.SERIALIZER) data = {"handshake": self._pyroHandshake} if config.METADATA: # the object id is only used/needed when piggybacking the metadata on the connection response # make sure to pass the resolved object id instead of the logical id data["object"] = uri.object flags = message.FLAGS_META_ON_CONNECT else: flags = 0 data, compressed = serializer.serializeData(data, config.COMPRESSION) if compressed: flags |= message.FLAGS_COMPRESSED msg = message.Message(message.MSG_CONNECT, data, serializer.serializer_id, flags, self._pyroSeq, annotations=self.__annotations(False), hmac_key=self._pyroHmacKey) if config.LOGWIRE: _log_wiredata(log, "proxy connect sending", msg) conn.send(msg.to_bytes()) msg = message.Message.recv(conn, [message.MSG_CONNECTOK, message.MSG_CONNECTFAIL], hmac_key=self._pyroHmacKey) if config.LOGWIRE: _log_wiredata(log, "proxy connect response received", msg) except Exception as x: if conn: conn.close() err = "cannot connect to %s: %s" % (connect_location, x) log.error(err) if isinstance(x, errors.CommunicationError): raise else: ce = errors.CommunicationError(err) if sys.version_info >= (3, 0): ce.__cause__ = x raise ce else: handshake_response = "?" if msg.data: serializer = util.get_serializer_by_id(msg.serializer_id) handshake_response = serializer.deserializeData(msg.data, compressed=msg.flags & message.FLAGS_COMPRESSED) if msg.type == message.MSG_CONNECTFAIL: if sys.version_info < (3, 0): error = "connection to %s rejected: %s" % (connect_location, handshake_response.decode()) else: error = "connection to %s rejected: %s" % (connect_location, handshake_response) conn.close() log.error(error) raise errors.CommunicationError(error) elif msg.type == message.MSG_CONNECTOK: if msg.flags & message.FLAGS_META_ON_CONNECT: self.__processMetadata(handshake_response["meta"]) handshake_response = handshake_response["handshake"] self._pyroConnection = conn if replaceUri: self._pyroUri = uri self._pyroValidateHandshake(handshake_response) log.debug("connected to %s - %s - %s", self._pyroUri, conn.family(), "SSL" if sslContext else "unencrypted") if msg.annotations: self._pyroResponseAnnotations(msg.annotations, msg.type) else: conn.close() err = "cannot connect to %s: invalid msg type %d received" % (connect_location, msg.type) log.error(err) raise errors.ProtocolError(err) with self.__pyroConnLock: if self._pyroConnection is not None: return False # already connected if connected_socket: if config.SSL and not isinstance(connected_socket, ssl.SSLSocket): raise socket.error("SSL configured for Pyro but existing socket is not a SSL socket") uri = self._pyroUri else: uri = _resolve(self._pyroUri, self._pyroHmacKey) # socket connection (normal or Unix domain socket) conn = None log.debug("connecting to %s", uri) connect_location = uri.sockname or (uri.host, uri.port) if connected_socket: self._pyroConnection = socketutil.SocketConnection(connected_socket, uri.object, True) else: connect_and_handshake(conn) if config.METADATA: # obtain metadata if this feature is enabled, and the metadata is not known yet if self._pyroMethods or self._pyroAttrs: log.debug("reusing existing metadata") else: self._pyroGetMetadata(uri.object) return True def _pyroGetMetadata(self, objectId=None, known_metadata=None): """ Get metadata from server (methods, attrs, oneway, ...) and remember them in some attributes of the proxy. Usually this will already be known due to the default behavior of the connect handshake, where the connect response also includes the metadata. """ objectId = objectId or self._pyroUri.object log.debug("getting metadata for object %s", objectId) if self._pyroConnection is None and not known_metadata: try: self.__pyroCreateConnection() except errors.PyroError: log.error("problem getting metadata: cannot connect") raise if self._pyroMethods or self._pyroAttrs: return # metadata has already been retrieved as part of creating the connection try: # invoke the get_metadata method on the daemon result = known_metadata or self._pyroInvoke("get_metadata", [objectId], {}, objectId=constants.DAEMON_NAME) self.__processMetadata(result) except errors.PyroError: log.exception("problem getting metadata") raise def __processMetadata(self, metadata): if not metadata: return self._pyroOneway = set(metadata["oneway"]) self._pyroMethods = set(metadata["methods"]) self._pyroAttrs = set(metadata["attrs"]) if log.isEnabledFor(logging.DEBUG): log.debug("from meta: methods=%s, oneway methods=%s, attributes=%s", sorted(self._pyroMethods), sorted(self._pyroOneway), sorted(self._pyroAttrs)) if not self._pyroMethods and not self._pyroAttrs: raise errors.PyroError("remote object doesn't expose any methods or attributes. Did you forget setting @expose on them?") def _pyroReconnect(self, tries=100000000): """ (Re)connect the proxy to the daemon containing the pyro object which the proxy is for. In contrast to the _pyroBind method, this one first releases the connection (if the proxy is still connected) and retries making a new connection until it succeeds or the given amount of tries ran out. """ self._pyroRelease() while tries: try: self.__pyroCreateConnection() return except errors.CommunicationError: tries -= 1 if tries: time.sleep(2) msg = "failed to reconnect" log.error(msg) raise errors.ConnectionClosedError(msg) def _pyroBatch(self): """returns a helper class that lets you create batched method calls on the proxy""" return _BatchProxyAdapter(self) def _pyroAsync(self, asynchronous=True): """turns the proxy into asynchronous mode so you can do asynchronous method calls, or sets it back to normal sync mode if you set asynchronous=False. This setting is strictly on a per-proxy basis (unless an exact clone is made via copy.copy).""" self.__async = asynchronous if sys.version_info < (3, 7): # async keyword backwards compatibility _pyroAsync_37 = _pyroAsync def _pyroAsync(self, asynchronous=True, **kwargs): if kwargs: kword = list(kwargs.keys()) if kword != ["async"]: raise TypeError("_pyroAsync() got an unexpected keyword argument '{:s}'".format(kword[0])) asynchronous = kwargs["async"] return Proxy._pyroAsync_37(self, asynchronous) def _pyroInvokeBatch(self, calls, oneway=False): flags = message.FLAGS_BATCH if oneway: flags |= message.FLAGS_ONEWAY return self._pyroInvoke("", calls, None, flags) def _pyroAnnotations(self): """ Override to return a dict with custom user annotations to be sent with each request message. Code using Pyro 4.56 or newer can skip this and instead set the annotations directly on the context object. """ return {} def _pyroResponseAnnotations(self, annotations, msgtype): """ Process any response annotations (dictionary set by the daemon). Usually this contains the internal Pyro annotations such as hmac and correlation id, and if you override the annotations method in the daemon, can contain your own annotations as well. Code using Pyro 4.56 or newer can skip this and instead read the response_annotations directly from the context object. """ pass def _pyroValidateHandshake(self, response): """ Process and validate the initial connection handshake response data received from the daemon. Simply return without error if everything is ok. Raise an exception if something is wrong and the connection should not be made. """ return def __annotations(self, clear=True): annotations = current_context.annotations if current_context.correlation_id: annotations["CORR"] = current_context.correlation_id.bytes else: annotations.pop("CORR", None) annotations.update(self._pyroAnnotations()) if clear: current_context.annotations = {} return annotations def __serializeBlobArgs(self, vargs, kwargs, annotations, flags, objectId, methodname, serializer): """ Special handling of a "blob" argument that has to stay serialized until explicitly deserialized in client code. This makes efficient, transparent gateways or dispatchers and such possible: they don't have to de/reserialize the message and are independent from the serialized class definitions. Annotations are passed in because some blob metadata is added. They're not part of the blob itself. """ if len(vargs) > 1 or kwargs: raise errors.SerializeError("if SerializedBlob is used, it must be the only argument") blob = vargs[0] flags |= message.FLAGS_KEEPSERIALIZED # Pass the objectId and methodname separately in an annotation because currently, # they are embedded inside the serialized message data. And we're not deserializing that, # so we have to have another means of knowing the object and method it is meant for... # A better solution is perhaps to split the actual remote method arguments from the # control data (object + methodname) but that requires a major protocol change. # The code below is not as nice but it works without any protocol change and doesn't # require a hack either - so it's actually not bad like this. import marshal annotations["BLBI"] = marshal.dumps((blob.info, objectId, methodname)) if blob._contains_blob: # directly pass through the already serialized msg data from within the blob protocol_msg = blob._data data, compressed = protocol_msg.data, protocol_msg.flags & message.FLAGS_COMPRESSED else: # replaces SerializedBlob argument with the data to be serialized data, compressed = serializer.serializeCall(objectId, methodname, blob._data, kwargs, compress=config.COMPRESSION) return data, compressed, flags class _StreamResultIterator(object): """ Pyro returns this as a result of a remote call which returns an iterator or generator. It is a normal iterable and produces elements on demand from the remote iterator. You can simply use it in for loops, list comprehensions etc. """ def __init__(self, streamId, proxy): self.streamId = streamId self.proxy = proxy self.pyroseq = proxy._pyroSeq def __iter__(self): return self def next(self): # python 2.x support return self.__next__() def __next__(self): if self.proxy is None: raise StopIteration if self.proxy._pyroConnection is None: raise errors.ConnectionClosedError("the proxy for this stream result has been closed") self.pyroseq += 1 try: return self.proxy._pyroInvoke("get_next_stream_item", [self.streamId], {}, objectId=constants.DAEMON_NAME) except (StopIteration, GeneratorExit): # when the iterator is exhausted, the proxy is removed to avoid unneeded close_stream calls later # (the server has closed its part of the stream by itself already) self.proxy = None raise def __del__(self): self.close() def close(self): if self.proxy and self.proxy._pyroConnection is not None: if self.pyroseq == self.proxy._pyroSeq: # we're still in sync, it's okay to use the same proxy to close this stream self.proxy._pyroInvoke("close_stream", [self.streamId], {}, flags=message.FLAGS_ONEWAY, objectId=constants.DAEMON_NAME) else: # The proxy's sequence number has diverged. # One of the reasons this can happen is because this call is being done from python's GC where # it decides to gc old iterator objects *during a new call on the proxy*. # If we use the same proxy and do a call in between, the other call on the proxy will get an out of sync seq and crash! # We create a temporary second proxy to call close_stream on. This is inefficient, but avoids the problem. try: with self.proxy.__copy__() as closingProxy: closingProxy._pyroInvoke("close_stream", [self.streamId], {}, flags=message.FLAGS_ONEWAY, objectId=constants.DAEMON_NAME) except errors.CommunicationError: pass self.proxy = None class _BatchedRemoteMethod(object): """method call abstraction that is used with batched calls""" def __init__(self, calls, name): self.__calls = calls self.__name = name def __getattr__(self, name): return _BatchedRemoteMethod(self.__calls, "%s.%s" % (self.__name, name)) def __call__(self, *args, **kwargs): self.__calls.append((self.__name, args, kwargs)) class _BatchProxyAdapter(object): """Helper class that lets you batch multiple method calls into one. It is constructed with a reference to the normal proxy that will carry out the batched calls. Call methods on this object that you want to batch, and finally call the batch proxy itself. That call will return a generator for the results of every method call in the batch (in sequence).""" def __init__(self, proxy): self.__proxy = proxy self.__calls = [] def __getattr__(self, name): return _BatchedRemoteMethod(self.__calls, name) def __enter__(self): return self def __exit__(self, *args): pass def __copy__(self): copy = type(self)(self.__proxy) copy.__calls = list(self.__calls) return copy def __resultsgenerator(self, results): for result in results: if isinstance(result, futures._ExceptionWrapper): result.raiseIt() # re-raise the remote exception locally. else: yield result # it is a regular result object, yield that and continue. def __call__(self, oneway=False, asynchronous=False): if oneway and asynchronous: raise errors.PyroError("async oneway calls make no sense") if asynchronous: return _AsyncRemoteMethod(self, "", self.__proxy._pyroMaxRetries)() else: results = self.__proxy._pyroInvokeBatch(self.__calls, oneway) self.__calls = [] # clear for re-use if not oneway: return self.__resultsgenerator(results) if sys.version_info < (3, 7): # async keyword backwards compatibility call_37 = __call__ def __call__(self, oneway=False, **kwargs): if kwargs: kword = list(kwargs.keys()) if kword != ["async"] and kword != ["asynchronous"]: raise TypeError("__call__() got an unexpected keyword argument '{:s}'".format(kword[0])) if kword == ["async"]: kwargs = {"asynchronous": kwargs["async"]} kwargs["oneway"] = oneway return _BatchProxyAdapter.call_37(self, **kwargs) def _pyroInvoke(self, name, args, kwargs): # ignore all parameters, we just need to execute the batch results = self.__proxy._pyroInvokeBatch(self.__calls) self.__calls = [] # clear for re-use return self.__resultsgenerator(results) class _AsyncRemoteMethod(object): """asynchronous method call abstraction (call will run in a background thread)""" def __init__(self, proxy, name, max_retries): self.__proxy = proxy self.__name = name self.__max_retries = max_retries def __getattr__(self, name): return _AsyncRemoteMethod(self.__proxy, "%s.%s" % (self.__name, name), self.__max_retries) def __call__(self, *args, **kwargs): result = futures.FutureResult() thread = threading.Thread(target=self.__asynccall, args=(result, args, kwargs)) thread.setDaemon(True) thread.start() return result def __asynccall(self, asyncresult, args, kwargs): for attempt in range(self.__max_retries + 1): try: # use a copy of the proxy otherwise calls would still be done in sequence, # and use contextmanager to close the proxy after we're done with self.__proxy.__copy__() as proxy: delay = 0.1 + random.random() / 5 while not proxy._pyroConnection: try: proxy._pyroBind() except errors.CommunicationError as x: if "no free workers" not in str(x): raise time.sleep(delay) # wait a bit until a worker might be available again delay += 0.4 + random.random() / 2 if 0 < config.COMMTIMEOUT / 2 < delay: raise value = proxy._pyroInvoke(self.__name, args, kwargs) asyncresult.value = value return except (errors.ConnectionClosedError, errors.TimeoutError) as x: # only retry for recoverable network errors if attempt >= self.__max_retries: # ignore any exceptions here, return them as part of the asynchronous result instead asyncresult.value = futures._ExceptionWrapper(x) return except Exception as x: # ignore any exceptions here, return them as part of the asynchronous result instead asyncresult.value = futures._ExceptionWrapper(x) return def batch(proxy): """convenience method to get a batch proxy adapter""" return proxy._pyroBatch() def asyncproxy(proxy, asynchronous=True): """convenience method to set proxy to asynchronous or sync mode.""" proxy._pyroAsync(asynchronous) def pyroObjectToAutoProxy(obj): """reduce function that automatically replaces Pyro objects by a Proxy""" if config.AUTOPROXY: daemon = getattr(obj, "_pyroDaemon", None) if daemon: # only return a proxy if the object is a registered pyro object return daemon.proxyFor(obj) return obj # decorators def callback(method): """ decorator to mark a method to be a 'callback'. This will make Pyro raise any errors also on the callback side, and not only on the side that does the callback call. """ method._pyroCallback = True return method def oneway(method): """ decorator to mark a method to be oneway (client won't wait for a response) """ method._pyroOneway = True return method def expose(method_or_class): """ Decorator to mark a method or class to be exposed for remote calls (relevant when REQUIRE_EXPOSE=True) You can apply it to a method or a class as a whole. If you need to change the default instance mode or instance creator, also use a @behavior decorator. """ if inspect.isdatadescriptor(method_or_class): func = method_or_class.fget or method_or_class.fset or method_or_class.fdel if util.is_private_attribute(func.__name__): raise AttributeError("exposing private names (starting with _) is not allowed") func._pyroExposed = True return method_or_class attrname = getattr(method_or_class, "__name__", None) if not attrname: # we could be dealing with a descriptor (classmethod/staticmethod), this means the order of the decorators is wrong if inspect.ismethoddescriptor(method_or_class): attrname = method_or_class.__get__(None, dict).__name__ raise AttributeError("using @expose on a classmethod/staticmethod must be done " "after @classmethod/@staticmethod. Method: " + attrname) else: raise AttributeError("@expose cannot determine what this is: " + repr(method_or_class)) if util.is_private_attribute(attrname): raise AttributeError("exposing private names (starting with _) is not allowed") if inspect.isclass(method_or_class): clazz = method_or_class log.debug("exposing all members of %r", clazz) for name in clazz.__dict__: if util.is_private_attribute(name): continue thing = getattr(clazz, name) if inspect.isfunction(thing) or inspect.ismethoddescriptor(thing): thing._pyroExposed = True elif inspect.ismethod(thing): thing.__func__._pyroExposed = True elif inspect.isdatadescriptor(thing): if getattr(thing, "fset", None): thing.fset._pyroExposed = True if getattr(thing, "fget", None): thing.fget._pyroExposed = True if getattr(thing, "fdel", None): thing.fdel._pyroExposed = True clazz._pyroExposed = True return clazz method_or_class._pyroExposed = True return method_or_class def behavior(instance_mode="session", instance_creator=None): """ Decorator to specify the server behavior of your Pyro class. """ def _behavior(clazz): if not inspect.isclass(clazz): raise TypeError("behavior decorator can only be used on a class") if instance_mode not in ("single", "session", "percall"): raise ValueError("invalid instance mode: " + instance_mode) if instance_creator and not callable(instance_creator): raise TypeError("instance_creator must be a callable") clazz._pyroInstancing = (instance_mode, instance_creator) return clazz if not isinstance(instance_mode, basestring): raise SyntaxError("behavior decorator is missing argument(s)") return _behavior @expose class DaemonObject(object): """The part of the daemon that is exposed as a Pyro object.""" def __init__(self, daemon): self.daemon = daemon def registered(self): """returns a list of all object names registered in this daemon""" return list(self.daemon.objectsById.keys()) def ping(self): """a simple do-nothing method for testing purposes""" pass def info(self): """return some descriptive information about the daemon""" return "%s bound on %s, NAT %s, %d objects registered. Servertype: %s" % ( constants.DAEMON_NAME, self.daemon.locationStr, self.daemon.natLocationStr, len(self.daemon.objectsById), self.daemon.transportServer) def get_metadata(self, objectId, as_lists=False): """ Get metadata for the given object (exposed methods, oneways, attributes). If you get an error in your proxy saying that 'DaemonObject' has no attribute 'get_metadata', you're probably connecting to an older Pyro version (4.26 or earlier). Either upgrade the Pyro version or set METADATA config item to False in your client code. """ obj = self.daemon.objectsById.get(objectId) if obj is not None: metadata = util.get_exposed_members(obj, only_exposed=config.REQUIRE_EXPOSE, as_lists=as_lists) if config.REQUIRE_EXPOSE and not metadata["methods"] and not metadata["attrs"]: # Something seems wrong: nothing is remotely exposed. # Possibly because older code not using @expose is now running with a more recent Pyro version # where @expose is mandatory in the default configuration. Give a hint to the user. if not inspect.isclass(obj): obj = type(obj) warnings.warn("Class %r doesn't expose any methods or attributes. Did you forget setting @expose on them?" % obj) return metadata else: log.debug("unknown object requested: %s", objectId) raise errors.DaemonError("unknown object") def get_next_stream_item(self, streamId): if streamId not in self.daemon.streaming_responses: raise errors.PyroError("item stream terminated") client, timestamp, linger_timestamp, stream = self.daemon.streaming_responses[streamId] if client is None: # reset client connection association (can be None if proxy disconnected) self.daemon.streaming_responses[streamId] = (current_context.client, timestamp, 0, stream) try: return next(stream) except Exception: # in case of error (or StopIteration!) the stream is removed del self.daemon.streaming_responses[streamId] raise def close_stream(self, streamId): if streamId in self.daemon.streaming_responses: del self.daemon.streaming_responses[streamId] class Daemon(object): """ Pyro daemon. Contains server side logic and dispatches incoming remote method calls to the appropriate objects. """ def __init__(self, host=None, port=0, unixsocket=None, nathost=None, natport=None, interface=DaemonObject, connected_socket=None): if connected_socket: nathost = natport = None else: if host is None: host = config.HOST if nathost is None: nathost = config.NATHOST if natport is None and nathost is not None: natport = config.NATPORT if nathost and unixsocket: raise ValueError("cannot use nathost together with unixsocket") if (nathost is None) ^ (natport is None): raise ValueError("must provide natport with nathost") self.__mustshutdown = threading.Event() self.__mustshutdown.set() self.__loopstopped = threading.Event() self.__loopstopped.set() if connected_socket: from Pyro4.socketserver.existingconnectionserver import SocketServer_ExistingConnection self.transportServer = SocketServer_ExistingConnection() self.transportServer.init(self, connected_socket) else: if config.SERVERTYPE == "thread": from Pyro4.socketserver.threadpoolserver import SocketServer_Threadpool self.transportServer = SocketServer_Threadpool() elif config.SERVERTYPE == "multiplex": from Pyro4.socketserver.multiplexserver import SocketServer_Multiplex self.transportServer = SocketServer_Multiplex() else: raise errors.PyroError("invalid server type '%s'" % config.SERVERTYPE) self.transportServer.init(self, host, port, unixsocket) #: The location (str of the form ``host:portnumber``) on which the Daemon is listening self.locationStr = self.transportServer.locationStr log.debug("daemon created on %s - %s (pid %d)", self.locationStr, socketutil.family_str(self.transportServer.sock), os.getpid()) natport_for_loc = natport if natport == 0: # expose internal port number as NAT port as well. (don't use port because it could be 0 and will be chosen by the OS) natport_for_loc = int(self.locationStr.split(":")[1]) #: The NAT-location (str of the form ``nathost:natportnumber``) on which the Daemon is exposed for use with NAT-routing self.natLocationStr = "%s:%d" % (nathost, natport_for_loc) if nathost else None if self.natLocationStr: log.debug("NAT address is %s", self.natLocationStr) pyroObject = interface(self) pyroObject._pyroId = constants.DAEMON_NAME #: Dictionary from Pyro object id to the actual Pyro object registered by this id self.objectsById = {pyroObject._pyroId: pyroObject} # assert that the configured serializers are available, and remember their ids: self.__serializer_ids = {util.get_serializer(ser_name).serializer_id for ser_name in config.SERIALIZERS_ACCEPTED} log.debug("accepted serializers: %s" % config.SERIALIZERS_ACCEPTED) log.debug("pyro protocol version: %d pickle version: %d" % (constants.PROTOCOL_VERSION, config.PICKLE_PROTOCOL_VERSION)) self.__pyroHmacKey = None self._pyroInstances = {} # pyro objects for instance_mode=single (singletons, just one per daemon) self.streaming_responses = {} # stream_id -> (client, creation_timestamp, linger_timestamp, stream) self.housekeeper_lock = threading.Lock() self.create_single_instance_lock = threading.Lock() self.__mustshutdown.clear() @property def _pyroHmacKey(self): return self.__pyroHmacKey @_pyroHmacKey.setter def _pyroHmacKey(self, value): # if needed, convert the hmac value to bytes first if value and sys.version_info >= (3, 0) and type(value) is not bytes: value = value.encode("utf-8") # convert to bytes self.__pyroHmacKey = value @property def sock(self): """the server socket used by the daemon""" return self.transportServer.sock @property def sockets(self): """list of all sockets used by the daemon (server socket and all active client sockets)""" return self.transportServer.sockets @property def selector(self): """the multiplexing selector used, if using the multiplex server type""" return self.transportServer.selector @staticmethod def serveSimple(objects, host=None, port=0, daemon=None, ns=True, verbose=True): """ Basic method to fire up a daemon (or supply one yourself). objects is a dict containing objects to register as keys, and their names (or None) as values. If ns is true they will be registered in the naming server as well, otherwise they just stay local. If you need to publish on a unix domain socket you can't use this shortcut method. See the documentation on 'publishing objects' (in chapter: Servers) for more details. """ if daemon is None: daemon = Daemon(host, port) with daemon: if ns: ns = _locateNS() for obj, name in objects.items(): if ns: localname = None # name is used for the name server else: localname = name # no name server, use name in daemon uri = daemon.register(obj, localname) if verbose: print("Object {0}:\n uri = {1}".format(repr(obj), uri)) if name and ns: ns.register(name, uri) if verbose: print(" name = {0}".format(name)) if verbose: print("Pyro daemon running.") daemon.requestLoop() def requestLoop(self, loopCondition=lambda: True): """ Goes in a loop to service incoming requests, until someone breaks this or calls shutdown from another thread. """ self.__mustshutdown.clear() log.info("daemon %s entering requestloop", self.locationStr) try: self.__loopstopped.clear() condition = lambda: not self.__mustshutdown.isSet() and loopCondition() self.transportServer.loop(loopCondition=condition) finally: self.__loopstopped.set() log.debug("daemon exits requestloop") def events(self, eventsockets): """for use in an external event loop: handle any requests that are pending for this daemon""" return self.transportServer.events(eventsockets) def shutdown(self): """Cleanly terminate a daemon that is running in the requestloop.""" log.debug("daemon shutting down") self.streaming_responses = {} time.sleep(0.02) self.__mustshutdown.set() if self.transportServer: self.transportServer.shutdown() time.sleep(0.02) self.close() self.__loopstopped.wait(timeout=5) # use timeout to avoid deadlock situations @property def _shutting_down(self): return self.__mustshutdown.is_set() def _handshake(self, conn, denied_reason=None): """ Perform connection handshake with new clients. Client sends a MSG_CONNECT message with a serialized data payload. If all is well, return with a CONNECT_OK message. The reason we're not doing this with a MSG_INVOKE method call on the daemon (like when retrieving the metadata) is because we need to force the clients to get past an initial connect handshake before letting them invoke any method. Return True for successful handshake, False if something was wrong. If a denied_reason is given, the handshake will fail with the given reason. """ serializer_id = util.MarshalSerializer.serializer_id msg_seq = 0 try: msg = message.Message.recv(conn, [message.MSG_CONNECT], hmac_key=self._pyroHmacKey) msg_seq = msg.seq if denied_reason: raise Exception(denied_reason) if config.LOGWIRE: _log_wiredata(log, "daemon handshake received", msg) if msg.serializer_id not in self.__serializer_ids: raise errors.SerializeError("message used serializer that is not accepted: %d" % msg.serializer_id) if "CORR" in msg.annotations: current_context.correlation_id = uuid.UUID(bytes=msg.annotations["CORR"]) else: current_context.correlation_id = uuid.uuid4() serializer_id = msg.serializer_id serializer = util.get_serializer_by_id(serializer_id) data = serializer.deserializeData(msg.data, msg.flags & message.FLAGS_COMPRESSED) handshake_response = self.validateHandshake(conn, data["handshake"]) if msg.flags & message.FLAGS_META_ON_CONNECT: # Usually this flag will be enabled, which results in including the object metadata # in the handshake response. This avoids a separate remote call to get_metadata. flags = message.FLAGS_META_ON_CONNECT handshake_response = { "handshake": handshake_response, "meta": self.objectsById[constants.DAEMON_NAME].get_metadata(data["object"], as_lists=True) } else: flags = 0 data, compressed = serializer.serializeData(handshake_response, config.COMPRESSION) msgtype = message.MSG_CONNECTOK if compressed: flags |= message.FLAGS_COMPRESSED except errors.ConnectionClosedError: log.debug("handshake failed, connection closed early") return False except Exception as x: log.debug("handshake failed, reason:", exc_info=True) serializer = util.get_serializer_by_id(serializer_id) data, compressed = serializer.serializeData(str(x), False) msgtype = message.MSG_CONNECTFAIL flags = message.FLAGS_COMPRESSED if compressed else 0 # We need a minimal amount of response data or the socket will remain blocked # on some systems... (messages smaller than 40 bytes) msg = message.Message(msgtype, data, serializer_id, flags, msg_seq, annotations=self.__annotations(), hmac_key=self._pyroHmacKey) if config.LOGWIRE: _log_wiredata(log, "daemon handshake response", msg) conn.send(msg.to_bytes()) return msg.type == message.MSG_CONNECTOK def validateHandshake(self, conn, data): """ Override this to create a connection validator for new client connections. It should return a response data object normally if the connection is okay, or should raise an exception if the connection should be denied. """ return "hello" def clientDisconnect(self, conn): """ Override this to handle a client disconnect. Conn is the SocketConnection object that was disconnected. """ pass def handleRequest(self, conn): """ Handle incoming Pyro request. Catches any exception that may occur and wraps it in a reply to the calling side, as to not make this server side loop terminate due to exceptions caused by remote invocations. """ request_flags = 0 request_seq = 0 request_serializer_id = util.MarshalSerializer.serializer_id wasBatched = False isCallback = False try: msg = message.Message.recv(conn, [message.MSG_INVOKE, message.MSG_PING], hmac_key=self._pyroHmacKey) except errors.CommunicationError as x: # we couldn't even get data from the client, this is an immediate error # log.info("error receiving data from client %s: %s", conn.sock.getpeername(), x) raise x try: request_flags = msg.flags request_seq = msg.seq request_serializer_id = msg.serializer_id current_context.correlation_id = uuid.UUID(bytes=msg.annotations["CORR"]) if "CORR" in msg.annotations else uuid.uuid4() if config.LOGWIRE: _log_wiredata(log, "daemon wiredata received", msg) if msg.type == message.MSG_PING: # return same seq, but ignore any data (it's a ping, not an echo). Nothing is deserialized. msg = message.Message(message.MSG_PING, b"pong", msg.serializer_id, 0, msg.seq, annotations=self.__annotations(), hmac_key=self._pyroHmacKey) if config.LOGWIRE: _log_wiredata(log, "daemon wiredata sending", msg) conn.send(msg.to_bytes()) return if msg.serializer_id not in self.__serializer_ids: raise errors.SerializeError("message used serializer that is not accepted: %d" % msg.serializer_id) serializer = util.get_serializer_by_id(msg.serializer_id) if request_flags & message.FLAGS_KEEPSERIALIZED: # pass on the wire protocol message blob unchanged objId, method, vargs, kwargs = self.__deserializeBlobArgs(msg) else: # normal deserialization of remote call arguments objId, method, vargs, kwargs = serializer.deserializeCall(msg.data, compressed=msg.flags & message.FLAGS_COMPRESSED) current_context.client = conn try: current_context.client_sock_addr = conn.sock.getpeername() # store, because on oneway calls, socket will be disconnected except socket.error: current_context.client_sock_addr = None # sometimes getpeername() doesn't work... current_context.seq = msg.seq current_context.annotations = msg.annotations current_context.msg_flags = msg.flags current_context.serializer_id = msg.serializer_id del msg # invite GC to collect the object, don't wait for out-of-scope obj = self.objectsById.get(objId) if obj is not None: if inspect.isclass(obj): obj = self._getInstance(obj, conn) if request_flags & message.FLAGS_BATCH: # batched method calls, loop over them all and collect all results data = [] for method, vargs, kwargs in vargs: method = util.getAttribute(obj, method) try: result = method(*vargs, **kwargs) # this is the actual method call to the Pyro object except Exception: xt, xv = sys.exc_info()[0:2] log.debug("Exception occurred while handling batched request: %s", xv) xv._pyroTraceback = util.formatTraceback(detailed=config.DETAILED_TRACEBACK) if sys.platform == "cli": util.fixIronPythonExceptionForPickle(xv, True) # piggyback attributes data.append(futures._ExceptionWrapper(xv)) break # stop processing the rest of the batch else: data.append(result) # note that we don't support streaming results in batch mode wasBatched = True else: # normal single method call if method == "__getattr__": # special case for direct attribute access (only exposed @properties are accessible) data = util.get_exposed_property_value(obj, vargs[0], only_exposed=config.REQUIRE_EXPOSE) elif method == "__setattr__": # special case for direct attribute access (only exposed @properties are accessible) data = util.set_exposed_property_value(obj, vargs[0], vargs[1], only_exposed=config.REQUIRE_EXPOSE) else: method = util.getAttribute(obj, method) if request_flags & message.FLAGS_ONEWAY and config.ONEWAY_THREADED: # oneway call to be run inside its own thread _OnewayCallThread(target=method, args=vargs, kwargs=kwargs).start() else: isCallback = getattr(method, "_pyroCallback", False) data = method(*vargs, **kwargs) # this is the actual method call to the Pyro object if not request_flags & message.FLAGS_ONEWAY: isStream, data = self._streamResponse(data, conn) if isStream: # throw an exception as well as setting message flags # this way, it is backwards compatible with older pyro versions. exc = errors.ProtocolError("result of call is an iterator") ann = {"STRM": data.encode()} if data else {} self._sendExceptionResponse(conn, request_seq, serializer.serializer_id, exc, None, annotations=ann, flags=message.FLAGS_ITEMSTREAMRESULT) return else: log.debug("unknown object requested: %s", objId) raise errors.DaemonError("unknown object") if request_flags & message.FLAGS_ONEWAY: return # oneway call, don't send a response else: data, compressed = serializer.serializeData(data, compress=config.COMPRESSION) response_flags = 0 if compressed: response_flags |= message.FLAGS_COMPRESSED if wasBatched: response_flags |= message.FLAGS_BATCH msg = message.Message(message.MSG_RESULT, data, serializer.serializer_id, response_flags, request_seq, annotations=self.__annotations(), hmac_key=self._pyroHmacKey) current_context.response_annotations = {} if config.LOGWIRE: _log_wiredata(log, "daemon wiredata sending", msg) conn.send(msg.to_bytes()) except Exception: xt, xv = sys.exc_info()[0:2] msg = getattr(xv, "pyroMsg", None) if msg: request_seq = msg.seq request_serializer_id = msg.serializer_id if xt is not errors.ConnectionClosedError: if xt not in (StopIteration, GeneratorExit): log.debug("Exception occurred while handling request: %r", xv) if not request_flags & message.FLAGS_ONEWAY: if isinstance(xv, errors.SerializeError) or not isinstance(xv, errors.CommunicationError): # only return the error to the client if it wasn't a oneway call, and not a communication error # (in these cases, it makes no sense to try to report the error back to the client...) tblines = util.formatTraceback(detailed=config.DETAILED_TRACEBACK) self._sendExceptionResponse(conn, request_seq, request_serializer_id, xv, tblines) if isCallback or isinstance(xv, (errors.CommunicationError, errors.SecurityError)): raise # re-raise if flagged as callback, communication or security error. def _clientDisconnect(self, conn): if config.ITER_STREAM_LINGER > 0: # client goes away, keep streams around for a bit longer (allow reconnect) for streamId in list(self.streaming_responses): info = self.streaming_responses.get(streamId, None) if info and info[0] is conn: _, timestamp, _, stream = info self.streaming_responses[streamId] = (None, timestamp, time.time(), stream) else: # client goes away, close any streams it had open as well for streamId in list(self.streaming_responses): info = self.streaming_responses.get(streamId, None) if info and info[0] is conn: del self.streaming_responses[streamId] self.clientDisconnect(conn) # user overridable hook def _housekeeping(self): """ Perform periodical housekeeping actions (cleanups etc) """ if self._shutting_down: return with self.housekeeper_lock: if self.streaming_responses: if config.ITER_STREAM_LIFETIME > 0: # cleanup iter streams that are past their lifetime for streamId in list(self.streaming_responses.keys()): info = self.streaming_responses.get(streamId, None) if info: last_use_period = time.time() - info[1] if 0 < config.ITER_STREAM_LIFETIME < last_use_period: del self.streaming_responses[streamId] if config.ITER_STREAM_LINGER > 0: # cleanup iter streams that are past their linger time for streamId in list(self.streaming_responses.keys()): info = self.streaming_responses.get(streamId, None) if info and info[2]: linger_period = time.time() - info[2] if linger_period > config.ITER_STREAM_LINGER: del self.streaming_responses[streamId] self.housekeeping() def housekeeping(self): """ Override this to add custom periodic housekeeping (cleanup) logic. This will be called every few seconds by the running daemon's request loop. """ pass def _getInstance(self, clazz, conn): """ Find or create a new instance of the class """ def createInstance(clazz, creator): try: if creator: obj = creator(clazz) if isinstance(obj, clazz): return obj raise TypeError("instance creator returned object of different type") return clazz() except Exception: log.exception("could not create pyro object instance") raise instance_mode, instance_creator = clazz._pyroInstancing if instance_mode == "single": # create and use one singleton instance of this class (not a global singleton, just exactly one per daemon) with self.create_single_instance_lock: instance = self._pyroInstances.get(clazz) if not instance: log.debug("instancemode %s: creating new pyro object for %s", instance_mode, clazz) instance = createInstance(clazz, instance_creator) self._pyroInstances[clazz] = instance return instance elif instance_mode == "session": # Create and use one instance for this proxy connection # the instances are kept on the connection object. # (this is the default instance mode when using new style @expose) instance = conn.pyroInstances.get(clazz) if not instance: log.debug("instancemode %s: creating new pyro object for %s", instance_mode, clazz) instance = createInstance(clazz, instance_creator) conn.pyroInstances[clazz] = instance return instance elif instance_mode == "percall": # create and use a new instance just for this call log.debug("instancemode %s: creating new pyro object for %s", instance_mode, clazz) return createInstance(clazz, instance_creator) else: raise errors.DaemonError("invalid instancemode in registered class") def _sendExceptionResponse(self, connection, seq, serializer_id, exc_value, tbinfo, flags=0, annotations=None): """send an exception back including the local traceback info""" exc_value._pyroTraceback = tbinfo if sys.platform == "cli": util.fixIronPythonExceptionForPickle(exc_value, True) # piggyback attributes serializer = util.get_serializer_by_id(serializer_id) try: data, compressed = serializer.serializeData(exc_value) except: # the exception object couldn't be serialized, use a generic PyroError instead xt, xv, tb = sys.exc_info() msg = "Error serializing exception: %s. Original exception: %s: %s" % (str(xv), type(exc_value), str(exc_value)) exc_value = errors.PyroError(msg) exc_value._pyroTraceback = tbinfo if sys.platform == "cli": util.fixIronPythonExceptionForPickle(exc_value, True) # piggyback attributes data, compressed = serializer.serializeData(exc_value) flags |= message.FLAGS_EXCEPTION if compressed: flags |= message.FLAGS_COMPRESSED annotations = dict(annotations or {}) annotations.update(self.annotations()) msg = message.Message(message.MSG_RESULT, data, serializer.serializer_id, flags, seq, annotations=annotations, hmac_key=self._pyroHmacKey) if config.LOGWIRE: _log_wiredata(log, "daemon wiredata sending (error response)", msg) connection.send(msg.to_bytes()) def register(self, obj_or_class, objectId=None, force=False): """ Register a Pyro object under the given id. Note that this object is now only known inside this daemon, it is not automatically available in a name server. This method returns a URI for the registered object. Pyro checks if an object is already registered, unless you set force=True. You can register a class or an object (instance) directly. For a class, Pyro will create instances of it to handle the remote calls according to the instance_mode (set via @expose on the class). The default there is one object per session (=proxy connection). If you register an object directly, Pyro will use that single object for *all* remote calls. """ if objectId: if not isinstance(objectId, basestring): raise TypeError("objectId must be a string or None") else: objectId = "obj_" + uuid.uuid4().hex # generate a new objectId if inspect.isclass(obj_or_class): if not hasattr(obj_or_class, "_pyroInstancing"): obj_or_class._pyroInstancing = ("session", None) if not force: if hasattr(obj_or_class, "_pyroId") and obj_or_class._pyroId != "": # check for empty string is needed for Cython raise errors.DaemonError("object or class already has a Pyro id") if objectId in self.objectsById: raise errors.DaemonError("an object or class is already registered with that id") # set some pyro attributes obj_or_class._pyroId = objectId obj_or_class._pyroDaemon = self if config.AUTOPROXY: # register a custom serializer for the type to automatically return proxies # we need to do this for all known serializers for ser in util._serializers.values(): if inspect.isclass(obj_or_class): ser.register_type_replacement(obj_or_class, pyroObjectToAutoProxy) else: ser.register_type_replacement(type(obj_or_class), pyroObjectToAutoProxy) # register the object/class in the mapping self.objectsById[obj_or_class._pyroId] = obj_or_class return self.uriFor(objectId) def unregister(self, objectOrId): """ Remove a class or object from the known objects inside this daemon. You can unregister the class/object directly, or with its id. """ if objectOrId is None: raise ValueError("object or objectid argument expected") if not isinstance(objectOrId, basestring): objectId = getattr(objectOrId, "_pyroId", None) if objectId is None: raise errors.DaemonError("object isn't registered") else: objectId = objectOrId objectOrId = None if objectId == constants.DAEMON_NAME: return if objectId in self.objectsById: del self.objectsById[objectId] if objectOrId is not None: del objectOrId._pyroId del objectOrId._pyroDaemon # Don't remove the custom type serializer because there may be # other registered objects of the same type still depending on it. def uriFor(self, objectOrId, nat=True): """ Get a URI for the given object (or object id) from this daemon. Only a daemon can hand out proper uris because the access location is contained in them. Note that unregistered objects cannot be given an uri, but unregistered object names can (it's just a string we're creating in that case). If nat is set to False, the configured NAT address (if any) is ignored and it will return an URI for the internal address. """ if not isinstance(objectOrId, basestring): objectOrId = getattr(objectOrId, "_pyroId", None) if objectOrId is None or objectOrId not in self.objectsById: raise errors.DaemonError("object isn't registered in this daemon") if nat: loc = self.natLocationStr or self.locationStr else: loc = self.locationStr return URI("PYRO:%s@%s" % (objectOrId, loc)) def resetMetadataCache(self, objectOrId, nat=True): """Reset cache of metadata when a Daemon has available methods/attributes dynamically updated. Clients will have to get a new proxy to see changes""" uri = self.uriFor(objectOrId, nat) # can only be cached if registered, else no-op if uri.object in self.objectsById: registered_object = self.objectsById[uri.object] # Clear cache regardless of how it is accessed util.reset_exposed_members(registered_object, config.REQUIRE_EXPOSE, as_lists=True) util.reset_exposed_members(registered_object, config.REQUIRE_EXPOSE, as_lists=False) def proxyFor(self, objectOrId, nat=True): """ Get a fully initialized Pyro Proxy for the given object (or object id) for this daemon. If nat is False, the configured NAT address (if any) is ignored. The object or id must be registered in this daemon, or you'll get an exception. (you can't get a proxy for an unknown object) """ uri = self.uriFor(objectOrId, nat) proxy = Proxy(uri) try: registered_object = self.objectsById[uri.object] except KeyError: raise errors.DaemonError("object isn't registered in this daemon") meta = util.get_exposed_members(registered_object, only_exposed=config.REQUIRE_EXPOSE) proxy._pyroGetMetadata(known_metadata=meta) return proxy def close(self): """Close down the server and release resources""" self.__mustshutdown.set() self.streaming_responses = {} if self.transportServer: log.debug("daemon closing") self.transportServer.close() self.transportServer = None def annotations(self): """Override to return a dict with custom user annotations to be sent with each response message.""" return {} def combine(self, daemon): """ Combines the event loop of the other daemon in the current daemon's loop. You can then simply run the current daemon's requestLoop to serve both daemons. This works fine on the multiplex server type, but doesn't work with the threaded server type. """ log.debug("combining event loop with other daemon") self.transportServer.combine_loop(daemon.transportServer) def __annotations(self): annotations = current_context.response_annotations if current_context.correlation_id: annotations["CORR"] = current_context.correlation_id.bytes else: annotations.pop("CORR", None) annotations.update(self.annotations()) return annotations def __repr__(self): if hasattr(self, "locationStr"): family = socketutil.family_str(self.sock) return "<%s.%s at 0x%x; %s - %s; %d objects>" % (self.__class__.__module__, self.__class__.__name__, id(self), self.locationStr, family, len(self.objectsById)) else: # daemon objects may come back from serialized form without being properly initialized (by design) return "<%s.%s at 0x%x; unusable>" % (self.__class__.__module__, self.__class__.__name__, id(self)) def __enter__(self): if not self.transportServer: raise errors.PyroError("cannot reuse this object") return self def __exit__(self, exc_type, exc_value, traceback): self.close() def __getstate__(self): # A little hack to make it possible to serialize Pyro objects, because they can reference a daemon, # but it is not meant to be able to properly serialize/deserialize Daemon objects. return {} def __getstate_for_dict__(self): return tuple(self.__getstate__()) def __setstate_from_dict__(self, state): pass if sys.version_info < (3, 0): __lazy_dict_iterator_types = (type({}.iterkeys()), type({}.itervalues()), type({}.iteritems())) else: __lazy_dict_iterator_types = (type({}.keys()), type({}.values()), type({}.items())) def _streamResponse(self, data, client): if sys.version_info < (3, 4): from collections import Iterator else: from collections.abc import Iterator if isinstance(data, Iterator) or inspect.isgenerator(data): if config.ITER_STREAMING: if type(data) in self.__lazy_dict_iterator_types: raise errors.PyroError("won't serialize or stream lazy dict iterators, convert to list yourself") stream_id = str(uuid.uuid4()) self.streaming_responses[stream_id] = (client, time.time(), 0, data) return True, stream_id return True, None return False, data def __deserializeBlobArgs(self, protocolmsg): import marshal blobinfo = protocolmsg.annotations["BLBI"] if sys.platform == "cli" and type(blobinfo) is not str: # Ironpython's marshal expects str... blobinfo = str(blobinfo) blobinfo, objId, method = marshal.loads(blobinfo) blob = SerializedBlob(blobinfo, protocolmsg, is_blob=True) return objId, method, (blob,), {} # object, method, vargs, kwargs # serpent serializer initialization try: import serpent def pyro_class_serpent_serializer(obj, serializer, stream, level): # Override the default way that a Pyro URI/proxy/daemon is serialized. # Because it defines a __getstate__ it would otherwise just become a tuple, # and not be deserialized as a class. d = util.SerializerBase.class_to_dict(obj) serializer.ser_builtins_dict(d, stream, level) # register the special serializers for the pyro objects with Serpent serpent.register_class(URI, pyro_class_serpent_serializer) serpent.register_class(Proxy, pyro_class_serpent_serializer) serpent.register_class(Daemon, pyro_class_serpent_serializer) serpent.register_class(futures._ExceptionWrapper, pyro_class_serpent_serializer) except ImportError: pass def serialize_core_object_to_dict(obj): return { "__class__": "Pyro4.core." + obj.__class__.__name__, "state": obj.__getstate_for_dict__() } util.SerializerBase.register_class_to_dict(URI, serialize_core_object_to_dict, serpent_too=False) util.SerializerBase.register_class_to_dict(Proxy, serialize_core_object_to_dict, serpent_too=False) util.SerializerBase.register_class_to_dict(Daemon, serialize_core_object_to_dict, serpent_too=False) util.SerializerBase.register_class_to_dict(futures._ExceptionWrapper, futures._ExceptionWrapper.__serialized_dict__, serpent_too=False) def _log_wiredata(logger, text, msg): """logs all the given properties of the wire message in the given logger""" corr = str(uuid.UUID(bytes=msg.annotations["CORR"])) if "CORR" in msg.annotations else "?" logger.debug("%s: msgtype=%d flags=0x%x ser=%d seq=%d corr=%s\nannotations=%r\ndata=%r" % (text, msg.type, msg.flags, msg.serializer_id, msg.seq, corr, msg.annotations, msg.data)) class _CallContext(threading.local): def __init__(self): # per-thread initialization self.client = None self.client_sock_addr = None self.seq = 0 self.msg_flags = 0 self.serializer_id = 0 self.annotations = {} self.response_annotations = {} self.correlation_id = None def to_global(self): if sys.platform != "cli": return dict(self.__dict__) # ironpython somehow has problems getting at the values, so do it manually: return { "client": self.client, "seq": self.seq, "msg_flags": self.msg_flags, "serializer_id": self.serializer_id, "annotations": self.annotations, "response_annotations": self.response_annotations, "correlation_id": self.correlation_id, "client_sock_addr": self.client_sock_addr } def from_global(self, values): self.client = values["client"] self.seq = values["seq"] self.msg_flags = values["msg_flags"] self.serializer_id = values["serializer_id"] self.annotations = values["annotations"] self.response_annotations = values["response_annotations"] self.correlation_id = values["correlation_id"] self.client_sock_addr = values["client_sock_addr"] def track_resource(self, resource): """keep a weak reference to the resource to be tracked for this connection""" if self.client: self.client.tracked_resources.add(resource) else: raise errors.PyroError("cannot track resource on a connectionless call") def untrack_resource(self, resource): """no longer track the resource for this connection""" if self.client: self.client.tracked_resources.discard(resource) else: raise errors.PyroError("cannot untrack resource on a connectionless call") class _OnewayCallThread(threading.Thread): def __init__(self, target, args, kwargs): super(_OnewayCallThread, self).__init__(target=target, args=args, kwargs=kwargs, name="oneway-call") self.daemon = True self.parent_context = current_context.to_global() def run(self): current_context.from_global(self.parent_context) super(_OnewayCallThread, self).run() # name server utility function, here to avoid cyclic dependencies def _resolve(uri, hmac_key=None): """ Resolve a 'magic' uri (PYRONAME, PYROMETA) into the direct PYRO uri. It finds a name server, and use that to resolve a PYRONAME uri into the direct PYRO uri pointing to the named object. If uri is already a PYRO uri, it is returned unmodified. You can consider this a shortcut function so that you don't have to locate and use a name server proxy yourself. Note: if you need to resolve more than a few names, consider using the name server directly instead of repeatedly calling this function, to avoid the name server lookup overhead from each call. """ if isinstance(uri, basestring): uri = URI(uri) elif not isinstance(uri, URI): raise TypeError("can only resolve Pyro URIs") if uri.protocol == "PYRO": return uri log.debug("resolving %s", uri) if uri.protocol == "PYRONAME": with _locateNS(uri.host, uri.port, hmac_key=hmac_key) as nameserver: return nameserver.lookup(uri.object) elif uri.protocol == "PYROMETA": with _locateNS(uri.host, uri.port, hmac_key=hmac_key) as nameserver: candidates = nameserver.list(metadata_all=uri.object) if candidates: candidate = random.choice(list(candidates.values())) log.debug("resolved to candidate %s", candidate) return URI(candidate) raise errors.NamingError("no registrations available with desired metadata properties %s" % uri.object) else: raise errors.PyroError("invalid uri protocol") # name server utility function, here to avoid cyclic dependencies def _locateNS(host=None, port=None, broadcast=True, hmac_key=None): """Get a proxy for a name server somewhere in the network.""" if host is None: # first try localhost if we have a good chance of finding it there if config.NS_HOST in ("localhost", "::1") or config.NS_HOST.startswith("127."): if ":" in config.NS_HOST: # ipv6 hosts = ["[%s]" % config.NS_HOST] else: # Some systems (Debian Linux) have 127.0.1.1 in the hosts file assigned to the hostname, # try this too for convenience sake (only if it's actually used as a valid ip address) try: socket.gethostbyaddr("127.0.1.1") hosts = [config.NS_HOST] if config.NS_HOST == "127.0.1.1" else [config.NS_HOST, "127.0.1.1"] except socket.error: hosts = [config.NS_HOST] for host in hosts: uristring = "PYRO:%s@%s:%d" % (constants.NAMESERVER_NAME, host, port or config.NS_PORT) log.debug("locating the NS: %s", uristring) proxy = Proxy(uristring) proxy._pyroHmacKey = hmac_key try: proxy._pyroBind() log.debug("located NS") return proxy except errors.PyroError: pass if config.PREFER_IP_VERSION == 6: broadcast = False # ipv6 doesn't have broadcast. We should probably use multicast.... if broadcast: # broadcast lookup if not port: port = config.NS_BCPORT log.debug("broadcast locate") sock = socketutil.createBroadcastSocket(reuseaddr=config.SOCK_REUSE, timeout=0.7) for _ in range(3): try: for bcaddr in config.parseAddressesString(config.BROADCAST_ADDRS): try: sock.sendto(b"GET_NSURI", 0, (bcaddr, port)) except socket.error as x: err = getattr(x, "errno", x.args[0]) # handle some errno's that some platforms like to throw: if err not in socketutil.ERRNO_EADDRNOTAVAIL and err not in socketutil.ERRNO_EADDRINUSE: raise data, _ = sock.recvfrom(100) sock.close() if sys.version_info >= (3, 0): data = data.decode("iso-8859-1") log.debug("located NS: %s", data) proxy = Proxy(data) proxy._pyroHmacKey = hmac_key return proxy except socket.timeout: continue try: sock.shutdown(socket.SHUT_RDWR) except (OSError, socket.error): pass sock.close() log.debug("broadcast locate failed, try direct connection on NS_HOST") else: log.debug("skipping broadcast lookup") # broadcast failed or skipped, try PYRO directly on specific host host = config.NS_HOST port = config.NS_PORT # pyro direct lookup if not port: port = config.NS_PORT if URI.isUnixsockLocation(host): uristring = "PYRO:%s@%s" % (constants.NAMESERVER_NAME, host) else: # if not a unix socket, check for ipv6 if ":" in host: host = "[%s]" % host uristring = "PYRO:%s@%s:%d" % (constants.NAMESERVER_NAME, host, port) uri = URI(uristring) log.debug("locating the NS: %s", uri) proxy = Proxy(uri) proxy._pyroHmacKey = hmac_key try: proxy._pyroBind() log.debug("located NS") return proxy except errors.PyroError as x: e = errors.NamingError("Failed to locate the nameserver") if sys.version_info >= (3, 0): e.__cause__ = x raise e class SerializedBlob(object): """ Used to wrap some data to make Pyro pass this object transparently (it keeps the serialized payload as-is) Only when you need to access the actual client data you can deserialize on demand. This makes efficient, transparent gateways or dispatchers and such possible: they don't have to de/reserialize the message and are independent from the serialized class definitions. You have to pass this as the only parameter to a remote method call for Pyro to understand it. Init arguments: ``info`` = some (small) descriptive data about the blob. Can be a simple id or name or guid. Must be marshallable. ``data`` = the actual client data payload that you want to transfer in the blob. Can be anything that you would otherwise have used as regular remote call arguments. """ def __init__(self, info, data, is_blob=False): self.info = info self._data = data self._contains_blob = is_blob def deserialized(self): """Retrieves the client data stored in this blob. Deserializes the data automatically if required.""" if self._contains_blob: protocol_msg = self._data serializer = util.get_serializer_by_id(protocol_msg.serializer_id) _, _, data, _ = serializer.deserializeData(protocol_msg.data, protocol_msg.flags & message.FLAGS_COMPRESSED) return data else: return self._data # call context thread local current_context = _CallContext() """the context object for the current call. (thread-local)""" # 'async' keyword backwards compatibility for Python versions older than 3.7. New code should not use this! if sys.version_info < (3, 7): def asyncproxy(proxy, asynchronous=True, **kwargs): """convenience method to set proxy to asynchronous or sync mode.""" if kwargs: kword = list(kwargs.keys()) if kword != ["async"]: raise TypeError("asyncproxy() got an unexpected keyword argument '{:s}'".format(kword[0])) asynchronous = kwargs["async"] proxy._pyroAsync(asynchronous) current_module = sys.modules[__name__] pyro4_module = __import__("Pyro4") current_module.__dict__["async"] = pyro4_module.__dict__["async"] = asyncproxy