Current File : //lib/python3/dist-packages/twisted/internet/abstract.py
# -*- test-case-name: twisted.test.test_abstract -*-
# Copyright (c) Twisted Matrix Laboratories.
# See LICENSE for details.
"""
Support for generic select()able objects.
"""
from socket import AF_INET, AF_INET6, inet_pton
from typing import Iterable, List, Optional
from zope.interface import implementer
from twisted.internet import interfaces, main
from twisted.python import failure, reflect
# Twisted Imports
from twisted.python.compat import lazyByteSlice
def _dataMustBeBytes(obj):
if not isinstance(obj, bytes): # no, really, I mean it
raise TypeError("Data must be bytes")
# Python 3.4+ can join bytes and memoryviews; using a
# memoryview prevents the slice from copying
def _concatenate(bObj, offset, bArray):
return b"".join([memoryview(bObj)[offset:]] + bArray)
class _ConsumerMixin:
"""
L{IConsumer} implementations can mix this in to get C{registerProducer} and
C{unregisterProducer} methods which take care of keeping track of a
producer's state.
Subclasses must provide three attributes which L{_ConsumerMixin} will read
but not write:
- connected: A C{bool} which is C{True} as long as the consumer has
someplace to send bytes (for example, a TCP connection), and then
C{False} when it no longer does.
- disconnecting: A C{bool} which is C{False} until something like
L{ITransport.loseConnection} is called, indicating that the send buffer
should be flushed and the connection lost afterwards. Afterwards,
C{True}.
- disconnected: A C{bool} which is C{False} until the consumer no longer
has a place to send bytes, then C{True}.
Subclasses must also override the C{startWriting} method.
@ivar producer: L{None} if no producer is registered, otherwise the
registered producer.
@ivar producerPaused: A flag indicating whether the producer is currently
paused.
@type producerPaused: L{bool}
@ivar streamingProducer: A flag indicating whether the producer was
registered as a streaming (ie push) producer or not (ie a pull
producer). This will determine whether the consumer may ever need to
pause and resume it, or if it can merely call C{resumeProducing} on it
when buffer space is available.
@ivar streamingProducer: C{bool} or C{int}
"""
producer = None
producerPaused = False
streamingProducer = False
def startWriting(self):
"""
Override in a subclass to cause the reactor to monitor this selectable
for write events. This will be called once in C{unregisterProducer} if
C{loseConnection} has previously been called, so that the connection can
actually close.
"""
raise NotImplementedError("%r did not implement startWriting")
def registerProducer(self, producer, streaming):
"""
Register to receive data from a producer.
This sets this selectable to be a consumer for a producer. When this
selectable runs out of data on a write() call, it will ask the producer
to resumeProducing(). When the FileDescriptor's internal data buffer is
filled, it will ask the producer to pauseProducing(). If the connection
is lost, FileDescriptor calls producer's stopProducing() method.
If streaming is true, the producer should provide the IPushProducer
interface. Otherwise, it is assumed that producer provides the
IPullProducer interface. In this case, the producer won't be asked to
pauseProducing(), but it has to be careful to write() data only when its
resumeProducing() method is called.
"""
if self.producer is not None:
raise RuntimeError(
"Cannot register producer %s, because producer %s was never "
"unregistered." % (producer, self.producer)
)
if self.disconnected:
producer.stopProducing()
else:
self.producer = producer
self.streamingProducer = streaming
if not streaming:
producer.resumeProducing()
def unregisterProducer(self):
"""
Stop consuming data from a producer, without disconnecting.
"""
self.producer = None
if self.connected and self.disconnecting:
self.startWriting()
@implementer(interfaces.ILoggingContext)
class _LogOwner:
"""
Mixin to help implement L{interfaces.ILoggingContext} for transports which
have a protocol, the log prefix of which should also appear in the
transport's log prefix.
"""
def _getLogPrefix(self, applicationObject: object) -> str:
"""
Determine the log prefix to use for messages related to
C{applicationObject}, which may or may not be an
L{interfaces.ILoggingContext} provider.
@return: A C{str} giving the log prefix to use.
"""
if interfaces.ILoggingContext.providedBy(applicationObject):
return applicationObject.logPrefix()
return applicationObject.__class__.__name__
def logPrefix(self):
"""
Override this method to insert custom logging behavior. Its
return value will be inserted in front of every line. It may
be called more times than the number of output lines.
"""
return "-"
@implementer(
interfaces.IPushProducer,
interfaces.IReadWriteDescriptor,
interfaces.IConsumer,
interfaces.ITransport,
interfaces.IHalfCloseableDescriptor,
)
class FileDescriptor(_ConsumerMixin, _LogOwner):
"""
An object which can be operated on by select().
This is an abstract superclass of all objects which may be notified when
they are readable or writable; e.g. they have a file-descriptor that is
valid to be passed to select(2).
"""
connected = 0
disconnected = 0
disconnecting = 0
_writeDisconnecting = False
_writeDisconnected = False
dataBuffer = b""
offset = 0
SEND_LIMIT = 128 * 1024
def __init__(self, reactor: Optional[interfaces.IReactorFDSet] = None):
"""
@param reactor: An L{IReactorFDSet} provider which this descriptor will
use to get readable and writeable event notifications. If no value
is given, the global reactor will be used.
"""
if not reactor:
from twisted.internet import reactor as _reactor
reactor = _reactor # type: ignore[assignment]
self.reactor = reactor
# will be added to dataBuffer in doWrite
self._tempDataBuffer: List[bytes] = []
self._tempDataLen = 0
def connectionLost(self, reason):
"""The connection was lost.
This is called when the connection on a selectable object has been
lost. It will be called whether the connection was closed explicitly,
an exception occurred in an event handler, or the other end of the
connection closed it first.
Clean up state here, but make sure to call back up to FileDescriptor.
"""
self.disconnected = 1
self.connected = 0
if self.producer is not None:
self.producer.stopProducing()
self.producer = None
self.stopReading()
self.stopWriting()
def writeSomeData(self, data: bytes) -> None:
"""
Write as much as possible of the given data, immediately.
This is called to invoke the lower-level writing functionality, such
as a socket's send() method, or a file's write(); this method
returns an integer or an exception. If an integer, it is the number
of bytes written (possibly zero); if an exception, it indicates the
connection was lost.
"""
raise NotImplementedError(
"%s does not implement writeSomeData" % reflect.qual(self.__class__)
)
def doRead(self):
"""
Called when data is available for reading.
Subclasses must override this method. The result will be interpreted
in the same way as a result of doWrite().
"""
raise NotImplementedError(
"%s does not implement doRead" % reflect.qual(self.__class__)
)
def doWrite(self):
"""
Called when data can be written.
@return: L{None} on success, an exception or a negative integer on
failure.
@see: L{twisted.internet.interfaces.IWriteDescriptor.doWrite}.
"""
if len(self.dataBuffer) - self.offset < self.SEND_LIMIT:
# If there is currently less than SEND_LIMIT bytes left to send
# in the string, extend it with the array data.
self.dataBuffer = _concatenate(
self.dataBuffer, self.offset, self._tempDataBuffer
)
self.offset = 0
self._tempDataBuffer = []
self._tempDataLen = 0
# Send as much data as you can.
if self.offset:
l = self.writeSomeData(lazyByteSlice(self.dataBuffer, self.offset))
else:
l = self.writeSomeData(self.dataBuffer)
# There is no writeSomeData implementation in Twisted which returns
# < 0, but the documentation for writeSomeData used to claim negative
# integers meant connection lost. Keep supporting this here,
# although it may be worth deprecating and removing at some point.
if isinstance(l, Exception) or l < 0:
return l
self.offset += l
# If there is nothing left to send,
if self.offset == len(self.dataBuffer) and not self._tempDataLen:
self.dataBuffer = b""
self.offset = 0
# stop writing.
self.stopWriting()
# If I've got a producer who is supposed to supply me with data,
if self.producer is not None and (
(not self.streamingProducer) or self.producerPaused
):
# tell them to supply some more.
self.producerPaused = False
self.producer.resumeProducing()
elif self.disconnecting:
# But if I was previously asked to let the connection die, do
# so.
return self._postLoseConnection()
elif self._writeDisconnecting:
# I was previously asked to half-close the connection. We
# set _writeDisconnected before calling handler, in case the
# handler calls loseConnection(), which will want to check for
# this attribute.
self._writeDisconnected = True
result = self._closeWriteConnection()
return result
return None
def _postLoseConnection(self):
"""Called after a loseConnection(), when all data has been written.
Whatever this returns is then returned by doWrite.
"""
# default implementation, telling reactor we're finished
return main.CONNECTION_DONE
def _closeWriteConnection(self):
# override in subclasses
pass
def writeConnectionLost(self, reason):
# in current code should never be called
self.connectionLost(reason)
def readConnectionLost(self, reason: failure.Failure) -> None:
# override in subclasses
self.connectionLost(reason)
def getHost(self):
# ITransport.getHost
raise NotImplementedError()
def getPeer(self):
# ITransport.getPeer
raise NotImplementedError()
def _isSendBufferFull(self):
"""
Determine whether the user-space send buffer for this transport is full
or not.
When the buffer contains more than C{self.bufferSize} bytes, it is
considered full. This might be improved by considering the size of the
kernel send buffer and how much of it is free.
@return: C{True} if it is full, C{False} otherwise.
"""
return len(self.dataBuffer) + self._tempDataLen > self.bufferSize
def _maybePauseProducer(self):
"""
Possibly pause a producer, if there is one and the send buffer is full.
"""
# If we are responsible for pausing our producer,
if self.producer is not None and self.streamingProducer:
# and our buffer is full,
if self._isSendBufferFull():
# pause it.
self.producerPaused = True
self.producer.pauseProducing()
def write(self, data: bytes) -> None:
"""Reliably write some data.
The data is buffered until the underlying file descriptor is ready
for writing. If there is more than C{self.bufferSize} data in the
buffer and this descriptor has a registered streaming producer, its
C{pauseProducing()} method will be called.
"""
_dataMustBeBytes(data)
if not self.connected or self._writeDisconnected:
return
if data:
self._tempDataBuffer.append(data)
self._tempDataLen += len(data)
self._maybePauseProducer()
self.startWriting()
def writeSequence(self, iovec: Iterable[bytes]) -> None:
"""
Reliably write a sequence of data.
Currently, this is a convenience method roughly equivalent to::
for chunk in iovec:
fd.write(chunk)
It may have a more efficient implementation at a later time or in a
different reactor.
As with the C{write()} method, if a buffer size limit is reached and a
streaming producer is registered, it will be paused until the buffered
data is written to the underlying file descriptor.
"""
for i in iovec:
_dataMustBeBytes(i)
if not self.connected or not iovec or self._writeDisconnected:
return
self._tempDataBuffer.extend(iovec)
for i in iovec:
self._tempDataLen += len(i)
self._maybePauseProducer()
self.startWriting()
def loseConnection(self):
"""Close the connection at the next available opportunity.
Call this to cause this FileDescriptor to lose its connection. It will
first write any data that it has buffered.
If there is data buffered yet to be written, this method will cause the
transport to lose its connection as soon as it's done flushing its
write buffer. If you have a producer registered, the connection won't
be closed until the producer is finished. Therefore, make sure you
unregister your producer when it's finished, or the connection will
never close.
"""
if self.connected and not self.disconnecting:
if self._writeDisconnected:
# doWrite won't trigger the connection close anymore
self.stopReading()
self.stopWriting()
self.connectionLost(failure.Failure(main.CONNECTION_DONE))
else:
self.stopReading()
self.startWriting()
self.disconnecting = 1
def loseWriteConnection(self):
self._writeDisconnecting = True
self.startWriting()
def stopReading(self):
"""Stop waiting for read availability.
Call this to remove this selectable from being notified when it is
ready for reading.
"""
self.reactor.removeReader(self)
def stopWriting(self):
"""Stop waiting for write availability.
Call this to remove this selectable from being notified when it is ready
for writing.
"""
self.reactor.removeWriter(self)
def startReading(self):
"""Start waiting for read availability."""
self.reactor.addReader(self)
def startWriting(self):
"""Start waiting for write availability.
Call this to have this FileDescriptor be notified whenever it is ready for
writing.
"""
self.reactor.addWriter(self)
# Producer/consumer implementation
# first, the consumer stuff. This requires no additional work, as
# any object you can write to can be a consumer, really.
producer = None
bufferSize = 2 ** 2 ** 2 ** 2
def stopConsuming(self):
"""Stop consuming data.
This is called when a producer has lost its connection, to tell the
consumer to go lose its connection (and break potential circular
references).
"""
self.unregisterProducer()
self.loseConnection()
# producer interface implementation
def resumeProducing(self):
if self.connected and not self.disconnecting:
self.startReading()
def pauseProducing(self):
self.stopReading()
def stopProducing(self):
self.loseConnection()
def fileno(self):
"""File Descriptor number for select().
This method must be overridden or assigned in subclasses to
indicate a valid file descriptor for the operating system.
"""
return -1
def isIPAddress(addr: str, family: int = AF_INET) -> bool:
"""
Determine whether the given string represents an IP address of the given
family; by default, an IPv4 address.
@param addr: A string which may or may not be the decimal dotted
representation of an IPv4 address.
@param family: The address family to test for; one of the C{AF_*} constants
from the L{socket} module. (This parameter has only been available
since Twisted 17.1.0; previously L{isIPAddress} could only test for IPv4
addresses.)
@return: C{True} if C{addr} represents an IPv4 address, C{False} otherwise.
"""
if isinstance(addr, bytes): # type: ignore[unreachable]
try: # type: ignore[unreachable]
addr = addr.decode("ascii")
except UnicodeDecodeError:
return False
if family == AF_INET6:
# On some platforms, inet_ntop fails unless the scope ID is valid; this
# is a test for whether the given string *is* an IP address, so strip
# any potential scope ID before checking.
addr = addr.split("%", 1)[0]
elif family == AF_INET:
# On Windows, where 3.5+ implement inet_pton, "0" is considered a valid
# IPv4 address, but we want to ensure we have all 4 segments.
if addr.count(".") != 3:
return False
else:
raise ValueError(f"unknown address family {family!r}")
try:
# This might be a native implementation or the one from
# twisted.python.compat.
inet_pton(family, addr)
except (ValueError, OSError):
return False
return True
def isIPv6Address(addr: str) -> bool:
"""
Determine whether the given string represents an IPv6 address.
@param addr: A string which may or may not be the hex
representation of an IPv6 address.
@type addr: C{str}
@return: C{True} if C{addr} represents an IPv6 address, C{False}
otherwise.
@rtype: C{bool}
"""
return isIPAddress(addr, AF_INET6)
__all__ = ["FileDescriptor", "isIPAddress", "isIPv6Address"]
Mini Shell