pyqtgraph/pyqtgraph/multiprocess/processes.py

524 lines
23 KiB
Python

# -*- coding: utf-8 -*-
import subprocess, atexit, os, sys, time, signal, inspect
import multiprocessing.connection
try:
import cPickle as pickle
except ImportError:
import pickle
from .remoteproxy import RemoteEventHandler, ClosedError, NoResultError, LocalObjectProxy, ObjectProxy
from ..Qt import QT_LIB, mkQApp
from ..util import cprint # color printing for debugging
__all__ = ['Process', 'QtProcess', 'ForkedProcess', 'ClosedError', 'NoResultError']
class Process(RemoteEventHandler):
"""
Bases: RemoteEventHandler
This class is used to spawn and control a new python interpreter.
It uses subprocess.Popen to start the new process and communicates with it
using multiprocessing.Connection objects over a network socket.
By default, the remote process will immediately enter an event-processing
loop that carries out requests send from the parent process.
Remote control works mainly through proxy objects::
proc = Process() ## starts process, returns handle
rsys = proc._import('sys') ## asks remote process to import 'sys', returns
## a proxy which references the imported module
rsys.stdout.write('hello\n') ## This message will be printed from the remote
## process. Proxy objects can usually be used
## exactly as regular objects are.
proc.close() ## Request the remote process shut down
Requests made via proxy objects may be synchronous or asynchronous and may
return objects either by proxy or by value (if they are picklable). See
ProxyObject for more information.
"""
_process_count = 1 # just used for assigning colors to each process for debugging
def __init__(self, name=None, target=None, executable=None, copySysPath=True, debug=False, timeout=20, wrapStdout=None, pyqtapis=None):
"""
============== =============================================================
**Arguments:**
name Optional name for this process used when printing messages
from the remote process.
target Optional function to call after starting remote process.
By default, this is startEventLoop(), which causes the remote
process to handle requests from the parent process until it
is asked to quit. If you wish to specify a different target,
it must be picklable (bound methods are not).
copySysPath If True, copy the contents of sys.path to the remote process.
If False, then only the path required to import pyqtgraph is
added.
debug If True, print detailed information about communication
with the child process.
wrapStdout If True (default on windows) then stdout and stderr from the
child process will be caught by the parent process and
forwarded to its stdout/stderr. This provides a workaround
for a python bug: http://bugs.python.org/issue3905
but has the side effect that child output is significantly
delayed relative to the parent output.
pyqtapis Optional dictionary of PyQt API version numbers to set before
importing pyqtgraph in the remote process.
============== =============================================================
"""
if target is None:
target = startEventLoop
if name is None:
name = str(self)
if executable is None:
executable = sys.executable
self.debug = 7 if debug is True else False # 7 causes printing in white
## random authentication key
authkey = os.urandom(20)
## Windows seems to have a hard time with hmac
if sys.platform.startswith('win'):
authkey = None
#print "key:", ' '.join([str(ord(x)) for x in authkey])
## Listen for connection from remote process (and find free port number)
l = multiprocessing.connection.Listener(('localhost', 0), authkey=authkey)
port = l.address[1]
## start remote process, instruct it to run target function
if copySysPath:
sysPath = sys.path
else:
# what path do we need to make target importable?
mod = inspect.getmodule(target)
modroot = sys.modules[mod.__name__.split('.')[0]]
sysPath = os.path.abspath(os.path.join(os.path.dirname(modroot.__file__), '..'))
bootstrap = os.path.abspath(os.path.join(os.path.dirname(__file__), 'bootstrap.py'))
self.debugMsg('Starting child process (%s %s)' % (executable, bootstrap))
# Decide on printing color for this process
if debug:
procDebug = (Process._process_count%6) + 1 # pick a color for this process to print in
Process._process_count += 1
else:
procDebug = False
if wrapStdout is None:
wrapStdout = sys.platform.startswith('win')
if wrapStdout:
## note: we need all three streams to have their own PIPE due to this bug:
## http://bugs.python.org/issue3905
stdout = subprocess.PIPE
stderr = subprocess.PIPE
self.proc = subprocess.Popen((executable, bootstrap), stdin=subprocess.PIPE, stdout=stdout, stderr=stderr)
## to circumvent the bug and still make the output visible, we use
## background threads to pass data from pipes to stdout/stderr
self._stdoutForwarder = FileForwarder(self.proc.stdout, "stdout", procDebug)
self._stderrForwarder = FileForwarder(self.proc.stderr, "stderr", procDebug)
else:
self.proc = subprocess.Popen((executable, bootstrap), stdin=subprocess.PIPE)
targetStr = pickle.dumps(target) ## double-pickle target so that child has a chance to
## set its sys.path properly before unpickling the target
pid = os.getpid() # we must send pid to child because windows does not have getppid
# When running in a venv on Windows platform, since Python >= 3.7.3, the launched
# subprocess is a grandchild instead of a child, leading to self.proc.pid not being
# the pid of the launched subprocess.
# https://bugs.python.org/issue38905
#
# As a workaround, when we detect such a situation, we perform exchange of pids via
# the multiprocessing connection. Technically, only the launched subprocess needs to
# send its pid back. Practically, we hijack the ppid parameter to indicate to the
# subprocess that pid exchange is needed.
xchg_pids = sys.platform == 'win32' and os.getenv('VIRTUAL_ENV') is not None
## Send everything the remote process needs to start correctly
data = dict(
name=name+'_child',
port=port,
authkey=authkey,
ppid=pid if not xchg_pids else None,
targetStr=targetStr,
path=sysPath,
qt_lib=QT_LIB,
debug=procDebug,
pyqtapis=pyqtapis,
)
pickle.dump(data, self.proc.stdin)
self.proc.stdin.close()
## open connection for remote process
self.debugMsg('Listening for child process on port %d, authkey=%s..' % (port, repr(authkey)))
while True:
try:
conn = l.accept()
break
except IOError as err:
if err.errno == 4: # interrupted; try again
continue
else:
raise
child_pid = self.proc.pid
if xchg_pids:
# corresponding code is in:
# remoteproxy.py::RemoteEventHandler.__init__()
conn.send(pid)
child_pid = conn.recv()
RemoteEventHandler.__init__(self, conn, name+'_parent', pid=child_pid, debug=self.debug)
self.debugMsg('Connected to child process.')
atexit.register(self.join)
def join(self, timeout=10):
self.debugMsg('Joining child process..')
if self.proc.poll() is None:
self.close()
start = time.time()
while self.proc.poll() is None:
if timeout is not None and time.time() - start > timeout:
raise Exception('Timed out waiting for remote process to end.')
time.sleep(0.05)
self.conn.close()
# Close remote polling threads, otherwise they will spin continuously
if hasattr(self, "_stdoutForwarder"):
self._stdoutForwarder.finish.set()
self._stderrForwarder.finish.set()
self._stdoutForwarder.join()
self._stderrForwarder.join()
self.debugMsg('Child process exited. (%d)' % self.proc.returncode)
def debugMsg(self, msg, *args):
if hasattr(self, '_stdoutForwarder'):
## Lock output from subprocess to make sure we do not get line collisions
with self._stdoutForwarder.lock:
with self._stderrForwarder.lock:
RemoteEventHandler.debugMsg(self, msg, *args)
else:
RemoteEventHandler.debugMsg(self, msg, *args)
def startEventLoop(name, port, authkey, ppid, debug=False):
if debug:
import os
cprint.cout(debug, '[%d] connecting to server at port localhost:%d, authkey=%s..\n'
% (os.getpid(), port, repr(authkey)), -1)
conn = multiprocessing.connection.Client(('localhost', int(port)), authkey=authkey)
if debug:
cprint.cout(debug, '[%d] connected; starting remote proxy.\n' % os.getpid(), -1)
global HANDLER
#ppid = 0 if not hasattr(os, 'getppid') else os.getppid()
HANDLER = RemoteEventHandler(conn, name, ppid, debug=debug)
while True:
try:
HANDLER.processRequests() # exception raised when the loop should exit
time.sleep(0.01)
except ClosedError:
HANDLER.debugMsg('Exiting server loop.')
sys.exit(0)
class ForkedProcess(RemoteEventHandler):
"""
ForkedProcess is a substitute for Process that uses os.fork() to generate a new process.
This is much faster than starting a completely new interpreter and child processes
automatically have a copy of the entire program state from before the fork. This
makes it an appealing approach when parallelizing expensive computations. (see
also Parallelizer)
However, fork() comes with some caveats and limitations:
- fork() is not available on Windows.
- It is not possible to have a QApplication in both parent and child process
(unless both QApplications are created _after_ the call to fork())
Attempts by the forked process to access Qt GUI elements created by the parent
will most likely cause the child to crash.
- Likewise, database connections are unlikely to function correctly in a forked child.
- Threads are not copied by fork(); the new process
will have only one thread that starts wherever fork() was called in the parent process.
- Forked processes are unceremoniously terminated when join() is called; they are not
given any opportunity to clean up. (This prevents them calling any cleanup code that
was only intended to be used by the parent process)
- Normally when fork()ing, open file handles are shared with the parent process,
which is potentially dangerous. ForkedProcess is careful to close all file handles
that are not explicitly needed--stdout, stderr, and a single pipe to the parent
process.
"""
def __init__(self, name=None, target=0, preProxy=None, randomReseed=True):
"""
When initializing, an optional target may be given.
If no target is specified, self.eventLoop will be used.
If None is given, no target will be called (and it will be up
to the caller to properly shut down the forked process)
preProxy may be a dict of values that will appear as ObjectProxy
in the remote process (but do not need to be sent explicitly since
they are available immediately before the call to fork().
Proxies will be availabe as self.proxies[name].
If randomReseed is True, the built-in random and numpy.random generators
will be reseeded in the child process.
"""
self.hasJoined = False
if target == 0:
target = self.eventLoop
if name is None:
name = str(self)
conn, remoteConn = multiprocessing.Pipe()
proxyIDs = {}
if preProxy is not None:
for k, v in preProxy.items():
proxyId = LocalObjectProxy.registerObject(v)
proxyIDs[k] = proxyId
ppid = os.getpid() # write this down now; windows doesn't have getppid
pid = os.fork()
if pid == 0:
self.isParent = False
## We are now in the forked process; need to be extra careful what we touch while here.
## - no reading/writing file handles/sockets owned by parent process (stdout is ok)
## - don't touch QtGui or QApplication at all; these are landmines.
## - don't let the process call exit handlers
os.setpgrp() ## prevents signals (notably keyboard interrupt) being forwarded from parent to this process
## close all file handles we do not want shared with parent
conn.close()
sys.stdin.close() ## otherwise we screw with interactive prompts.
fid = remoteConn.fileno()
os.closerange(3, fid)
os.closerange(fid+1, 4096) ## just guessing on the maximum descriptor count..
## Override any custom exception hooks
def excepthook(*args):
import traceback
traceback.print_exception(*args)
sys.excepthook = excepthook
## Make it harder to access QApplication instance
for qtlib in ('PyQt4', 'PySide', 'PyQt5'):
if qtlib in sys.modules:
sys.modules[qtlib+'.QtGui'].QApplication = None
sys.modules.pop(qtlib+'.QtGui', None)
sys.modules.pop(qtlib+'.QtCore', None)
## sabotage atexit callbacks
atexit._exithandlers = []
atexit.register(lambda: os._exit(0))
if randomReseed:
if 'numpy.random' in sys.modules:
sys.modules['numpy.random'].seed(os.getpid() ^ int(time.time()*10000%10000))
if 'random' in sys.modules:
sys.modules['random'].seed(os.getpid() ^ int(time.time()*10000%10000))
#ppid = 0 if not hasattr(os, 'getppid') else os.getppid()
RemoteEventHandler.__init__(self, remoteConn, name+'_child', pid=ppid)
self.forkedProxies = {}
for name, proxyId in proxyIDs.items():
self.forkedProxies[name] = ObjectProxy(ppid, proxyId=proxyId, typeStr=repr(preProxy[name]))
if target is not None:
target()
else:
self.isParent = True
self.childPid = pid
remoteConn.close()
RemoteEventHandler.handlers = {} ## don't want to inherit any of this from the parent.
RemoteEventHandler.__init__(self, conn, name+'_parent', pid=pid)
atexit.register(self.join)
def eventLoop(self):
while True:
try:
self.processRequests() # exception raised when the loop should exit
time.sleep(0.01)
except ClosedError:
break
except:
print("Error occurred in forked event loop:")
sys.excepthook(*sys.exc_info())
sys.exit(0)
def join(self, timeout=10):
if self.hasJoined:
return
#os.kill(pid, 9)
try:
self.close(callSync='sync', timeout=timeout, noCleanup=True) ## ask the child process to exit and require that it return a confirmation.
except IOError: ## probably remote process has already quit
pass
try:
os.waitpid(self.childPid, 0)
except OSError: ## probably remote process has already quit
pass
self.conn.close() # don't leak file handles!
self.hasJoined = True
def kill(self):
"""Immediately kill the forked remote process.
This is generally safe because forked processes are already
expected to _avoid_ any cleanup at exit."""
os.kill(self.childPid, signal.SIGKILL)
self.hasJoined = True
##Special set of subclasses that implement a Qt event loop instead.
class RemoteQtEventHandler(RemoteEventHandler):
def __init__(self, *args, **kwds):
RemoteEventHandler.__init__(self, *args, **kwds)
def startEventTimer(self):
from ..Qt import QtCore
self.timer = QtCore.QTimer()
self.timer.timeout.connect(self.processRequests)
self.timer.start(10)
def processRequests(self):
try:
RemoteEventHandler.processRequests(self)
except ClosedError:
from ..Qt import QtGui
QtGui.QApplication.instance().quit()
self.timer.stop()
#raise SystemExit
class QtProcess(Process):
"""
QtProcess is essentially the same as Process, with two major differences:
- The remote process starts by running startQtEventLoop() which creates a
QApplication in the remote process and uses a QTimer to trigger
remote event processing. This allows the remote process to have its own
GUI.
- A QTimer is also started on the parent process which polls for requests
from the child process. This allows Qt signals emitted within the child
process to invoke slots on the parent process and vice-versa. This can
be disabled using processRequests=False in the constructor.
Example::
proc = QtProcess()
rQtGui = proc._import('PyQt4.QtGui')
btn = rQtGui.QPushButton('button on child process')
btn.show()
def slot():
print('slot invoked on parent process')
btn.clicked.connect(proxy(slot)) # be sure to send a proxy of the slot
"""
def __init__(self, **kwds):
if 'target' not in kwds:
kwds['target'] = startQtEventLoop
from ..Qt import QtGui ## avoid module-level import to keep bootstrap snappy.
self._processRequests = kwds.pop('processRequests', True)
if self._processRequests and QtGui.QApplication.instance() is None:
raise Exception("Must create QApplication before starting QtProcess, or use QtProcess(processRequests=False)")
Process.__init__(self, **kwds)
self.startEventTimer()
def startEventTimer(self):
from ..Qt import QtCore ## avoid module-level import to keep bootstrap snappy.
self.timer = QtCore.QTimer()
if self._processRequests:
self.startRequestProcessing()
def startRequestProcessing(self, interval=0.01):
"""Start listening for requests coming from the child process.
This allows signals to be connected from the child process to the parent.
"""
self.timer.timeout.connect(self.processRequests)
self.timer.start(int(interval*1000))
def stopRequestProcessing(self):
self.timer.stop()
def processRequests(self):
try:
Process.processRequests(self)
except ClosedError:
self.timer.stop()
def startQtEventLoop(name, port, authkey, ppid, debug=False):
if debug:
import os
cprint.cout(debug, '[%d] connecting to server at port localhost:%d, authkey=%s..\n' % (os.getpid(), port, repr(authkey)), -1)
conn = multiprocessing.connection.Client(('localhost', int(port)), authkey=authkey)
if debug:
cprint.cout(debug, '[%d] connected; starting remote proxy.\n' % os.getpid(), -1)
from ..Qt import QtGui
app = QtGui.QApplication.instance()
#print app
if app is None:
app = mkQApp()
app.setQuitOnLastWindowClosed(False) ## generally we want the event loop to stay open
## until it is explicitly closed by the parent process.
global HANDLER
HANDLER = RemoteQtEventHandler(conn, name, ppid, debug=debug)
HANDLER.startEventTimer()
app.exec() if hasattr(app, 'exec') else app.exec_()
import threading
class FileForwarder(threading.Thread):
"""
Background thread that forwards data from one pipe to another.
This is used to catch data from stdout/stderr of the child process
and print it back out to stdout/stderr. We need this because this
bug: http://bugs.python.org/issue3905 _requires_ us to catch
stdout/stderr.
*output* may be a file or 'stdout' or 'stderr'. In the latter cases,
sys.stdout/stderr are retrieved once for every line that is output,
which ensures that the correct behavior is achieved even if
sys.stdout/stderr are replaced at runtime.
"""
def __init__(self, input, output, color):
threading.Thread.__init__(self)
self.input = input
self.output = output
self.lock = threading.Lock()
self.daemon = True
self.color = color
self.finish = threading.Event()
self.start()
def run(self):
if self.output == 'stdout' and self.color is not False:
while not self.finish.is_set():
line = self.input.readline()
with self.lock:
cprint.cout(self.color, line.decode('utf8'), -1)
elif self.output == 'stderr' and self.color is not False:
while not self.finish.is_set():
line = self.input.readline()
with self.lock:
cprint.cerr(self.color, line.decode('utf8'), -1)
else:
if isinstance(self.output, str):
self.output = getattr(sys, self.output)
while not self.finish.is_set():
line = self.input.readline()
with self.lock:
self.output.write(line.decode('utf8'))