Related
The function foo below returns a string 'foo'. How can I get the value 'foo' which is returned from the thread's target?
from threading import Thread
def foo(bar):
print('hello {}'.format(bar))
return 'foo'
thread = Thread(target=foo, args=('world!',))
thread.start()
return_value = thread.join()
The "one obvious way to do it", shown above, doesn't work: thread.join() returned None.
One way I've seen is to pass a mutable object, such as a list or a dictionary, to the thread's constructor, along with a an index or other identifier of some sort. The thread can then store its results in its dedicated slot in that object. For example:
def foo(bar, result, index):
print 'hello {0}'.format(bar)
result[index] = "foo"
from threading import Thread
threads = [None] * 10
results = [None] * 10
for i in range(len(threads)):
threads[i] = Thread(target=foo, args=('world!', results, i))
threads[i].start()
# do some other stuff
for i in range(len(threads)):
threads[i].join()
print " ".join(results) # what sound does a metasyntactic locomotive make?
If you really want join() to return the return value of the called function, you can do this with a Thread subclass like the following:
from threading import Thread
def foo(bar):
print 'hello {0}'.format(bar)
return "foo"
class ThreadWithReturnValue(Thread):
def __init__(self, group=None, target=None, name=None,
args=(), kwargs={}, Verbose=None):
Thread.__init__(self, group, target, name, args, kwargs, Verbose)
self._return = None
def run(self):
if self._Thread__target is not None:
self._return = self._Thread__target(*self._Thread__args,
**self._Thread__kwargs)
def join(self):
Thread.join(self)
return self._return
twrv = ThreadWithReturnValue(target=foo, args=('world!',))
twrv.start()
print twrv.join() # prints foo
That gets a little hairy because of some name mangling, and it accesses "private" data structures that are specific to Thread implementation... but it works.
For Python 3:
class ThreadWithReturnValue(Thread):
def __init__(self, group=None, target=None, name=None,
args=(), kwargs={}, Verbose=None):
Thread.__init__(self, group, target, name, args, kwargs)
self._return = None
def run(self):
if self._target is not None:
self._return = self._target(*self._args,
**self._kwargs)
def join(self, *args):
Thread.join(self, *args)
return self._return
FWIW, the multiprocessing module has a nice interface for this using the Pool class. And if you want to stick with threads rather than processes, you can just use the multiprocessing.pool.ThreadPool class as a drop-in replacement.
def foo(bar, baz):
print 'hello {0}'.format(bar)
return 'foo' + baz
from multiprocessing.pool import ThreadPool
pool = ThreadPool(processes=1)
async_result = pool.apply_async(foo, ('world', 'foo')) # tuple of args for foo
# do some other stuff in the main process
return_val = async_result.get() # get the return value from your function.
In Python 3.2+, stdlib concurrent.futures module provides a higher level API to threading, including passing return values or exceptions from a worker thread back to the main thread:
import concurrent.futures
def foo(bar):
print('hello {}'.format(bar))
return 'foo'
with concurrent.futures.ThreadPoolExecutor() as executor:
future = executor.submit(foo, 'world!')
return_value = future.result()
print(return_value)
Jake's answer is good, but if you don't want to use a threadpool (you don't know how many threads you'll need, but create them as needed) then a good way to transmit information between threads is the built-in Queue.Queue class, as it offers thread safety.
I created the following decorator to make it act in a similar fashion to the threadpool:
def threaded(f, daemon=False):
import Queue
def wrapped_f(q, *args, **kwargs):
'''this function calls the decorated function and puts the
result in a queue'''
ret = f(*args, **kwargs)
q.put(ret)
def wrap(*args, **kwargs):
'''this is the function returned from the decorator. It fires off
wrapped_f in a new thread and returns the thread object with
the result queue attached'''
q = Queue.Queue()
t = threading.Thread(target=wrapped_f, args=(q,)+args, kwargs=kwargs)
t.daemon = daemon
t.start()
t.result_queue = q
return t
return wrap
Then you just use it as:
#threaded
def long_task(x):
import time
x = x + 5
time.sleep(5)
return x
# does not block, returns Thread object
y = long_task(10)
print y
# this blocks, waiting for the result
result = y.result_queue.get()
print result
The decorated function creates a new thread each time it's called and returns a Thread object that contains the queue that will receive the result.
UPDATE
It's been quite a while since I posted this answer, but it still gets views so I thought I would update it to reflect the way I do this in newer versions of Python:
Python 3.2 added in the concurrent.futures module which provides a high-level interface for parallel tasks. It provides ThreadPoolExecutor and ProcessPoolExecutor, so you can use a thread or process pool with the same api.
One benefit of this api is that submitting a task to an Executor returns a Future object, which will complete with the return value of the callable you submit.
This makes attaching a queue object unnecessary, which simplifies the decorator quite a bit:
_DEFAULT_POOL = ThreadPoolExecutor()
def threadpool(f, executor=None):
#wraps(f)
def wrap(*args, **kwargs):
return (executor or _DEFAULT_POOL).submit(f, *args, **kwargs)
return wrap
This will use a default module threadpool executor if one is not passed in.
The usage is very similar to before:
#threadpool
def long_task(x):
import time
x = x + 5
time.sleep(5)
return x
# does not block, returns Future object
y = long_task(10)
print y
# this blocks, waiting for the result
result = y.result()
print result
If you're using Python 3.4+, one really nice feature of using this method (and Future objects in general) is that the returned future can be wrapped to turn it into an asyncio.Future with asyncio.wrap_future. This makes it work easily with coroutines:
result = await asyncio.wrap_future(long_task(10))
If you don't need access to the underlying concurrent.Future object, you can include the wrap in the decorator:
_DEFAULT_POOL = ThreadPoolExecutor()
def threadpool(f, executor=None):
#wraps(f)
def wrap(*args, **kwargs):
return asyncio.wrap_future((executor or _DEFAULT_POOL).submit(f, *args, **kwargs))
return wrap
Then, whenever you need to push cpu intensive or blocking code off the event loop thread, you can put it in a decorated function:
#threadpool
def some_long_calculation():
...
# this will suspend while the function is executed on a threadpool
result = await some_long_calculation()
Another solution that doesn't require changing your existing code:
import Queue # Python 2.x
#from queue import Queue # Python 3.x
from threading import Thread
def foo(bar):
print 'hello {0}'.format(bar) # Python 2.x
#print('hello {0}'.format(bar)) # Python 3.x
return 'foo'
que = Queue.Queue() # Python 2.x
#que = Queue() # Python 3.x
t = Thread(target=lambda q, arg1: q.put(foo(arg1)), args=(que, 'world!'))
t.start()
t.join()
result = que.get()
print result # Python 2.x
#print(result) # Python 3.x
It can be also easily adjusted to a multi-threaded environment:
import Queue # Python 2.x
#from queue import Queue # Python 3.x
from threading import Thread
def foo(bar):
print 'hello {0}'.format(bar) # Python 2.x
#print('hello {0}'.format(bar)) # Python 3.x
return 'foo'
que = Queue.Queue() # Python 2.x
#que = Queue() # Python 3.x
threads_list = list()
t = Thread(target=lambda q, arg1: q.put(foo(arg1)), args=(que, 'world!'))
t.start()
threads_list.append(t)
# Add more threads here
...
threads_list.append(t2)
...
threads_list.append(t3)
...
# Join all the threads
for t in threads_list:
t.join()
# Check thread's return value
while not que.empty():
result = que.get()
print result # Python 2.x
#print(result) # Python 3.x
UPDATE:
I think there's a significantly simpler and more concise way to save the result of the thread, and in a way that keeps the interface virtually identical to the threading.Thread class (please let me know if there are edge cases - I haven't tested as much as my original post below):
import threading
class ConciseResult(threading.Thread):
def run(self):
self.result = self._target(*self._args, **self._kwargs)
To be robust and avoid potential errors:
import threading
class ConciseRobustResult(threading.Thread):
def run(self):
try:
if self._target is not None:
self.result = self._target(*self._args, **self._kwargs)
finally:
# Avoid a refcycle if the thread is running a function with
# an argument that has a member that points to the thread.
del self._target, self._args, self._kwargs
Short explanation: we override only the run method of threading.Thread, and modify nothing else. This allows us to use everything else the threading.Thread class does for us, without needing to worry about missing potential edge cases such as _private attribute assignments or custom attribute modifications in the way that my original post does.
We can verify that we only modify the run method by looking at the output of help(ConciseResult) and help(ConciseRobustResult). The only method/attribute/descriptor included under Methods defined here: is run, and everything else comes from the inherited threading.Thread base class (see the Methods inherited from threading.Thread: section).
To test either of these implementations using the example code below, substitute ConciseResult or ConciseRobustResult for ThreadWithResult in the main function below.
Original post using a closure function in the init method:
Most answers I've found are long and require being familiar with other modules or advanced python features, and will be rather confusing to someone unless they're already familiar with everything the answer talks about.
Working code for a simplified approach:
import threading
class ThreadWithResult(threading.Thread):
def __init__(self, group=None, target=None, name=None, args=(), kwargs={}, *, daemon=None):
def function():
self.result = target(*args, **kwargs)
super().__init__(group=group, target=function, name=name, daemon=daemon)
Example code:
import time, random
def function_to_thread(n):
count = 0
while count < 3:
print(f'still running thread {n}')
count +=1
time.sleep(3)
result = random.random()
print(f'Return value of thread {n} should be: {result}')
return result
def main():
thread1 = ThreadWithResult(target=function_to_thread, args=(1,))
thread2 = ThreadWithResult(target=function_to_thread, args=(2,))
thread1.start()
thread2.start()
thread1.join()
thread2.join()
print(thread1.result)
print(thread2.result)
main()
Explanation:
I wanted to simplify things significantly, so I created a ThreadWithResult class and had it inherit from threading.Thread. The nested function function in __init__ calls the threaded function we want to save the value of, and saves the result of that nested function as the instance attribute self.result after the thread finishes executing.
Creating an instance of this is identical to creating an instance of threading.Thread. Pass in the function you want to run on a new thread to the target argument and any arguments that your function might need to the args argument and any keyword arguments to the kwargs argument.
e.g.
my_thread = ThreadWithResult(target=my_function, args=(arg1, arg2, arg3))
I think this is significantly easier to understand than the vast majority of answers, and this approach requires no extra imports! I included the time and random module to simulate the behavior of a thread, but they're not required to achieve the functionality asked in the original question.
I know I'm answering this looong after the question was asked, but I hope this can help more people in the future!
EDIT: I created the save-thread-result PyPI package to allow you to access the same code above and reuse it across projects (GitHub code is here). The PyPI package fully extends the threading.Thread class, so you can set any attributes you would set on threading.thread on the ThreadWithResult class as well!
The original answer above goes over the main idea behind this subclass, but for more information, see the more detailed explanation (from the module docstring) here.
Quick usage example:
pip3 install -U save-thread-result # MacOS/Linux
pip install -U save-thread-result # Windows
python3 # MacOS/Linux
python # Windows
from save_thread_result import ThreadWithResult
# As of Release 0.0.3, you can also specify values for
#`group`, `name`, and `daemon` if you want to set those
# values manually.
thread = ThreadWithResult(
target = my_function,
args = (my_function_arg1, my_function_arg2, ...)
kwargs = {my_function_kwarg1: kwarg1_value, my_function_kwarg2: kwarg2_value, ...}
)
thread.start()
thread.join()
if getattr(thread, 'result', None):
print(thread.result)
else:
# thread.result attribute not set - something caused
# the thread to terminate BEFORE the thread finished
# executing the function passed in through the
# `target` argument
print('ERROR! Something went wrong while executing this thread, and the function you passed in did NOT complete!!')
# seeing help about the class and information about the threading.Thread super class methods and attributes available:
help(ThreadWithResult)
Parris / kindall's answer join/return answer ported to Python 3:
from threading import Thread
def foo(bar):
print('hello {0}'.format(bar))
return "foo"
class ThreadWithReturnValue(Thread):
def __init__(self, group=None, target=None, name=None, args=(), kwargs=None, *, daemon=None):
Thread.__init__(self, group, target, name, args, kwargs, daemon=daemon)
self._return = None
def run(self):
if self._target is not None:
self._return = self._target(*self._args, **self._kwargs)
def join(self):
Thread.join(self)
return self._return
twrv = ThreadWithReturnValue(target=foo, args=('world!',))
twrv.start()
print(twrv.join()) # prints foo
Note, the Thread class is implemented differently in Python 3.
I stole kindall's answer and cleaned it up just a little bit.
The key part is adding *args and **kwargs to join() in order to handle the timeout
class threadWithReturn(Thread):
def __init__(self, *args, **kwargs):
super(threadWithReturn, self).__init__(*args, **kwargs)
self._return = None
def run(self):
if self._Thread__target is not None:
self._return = self._Thread__target(*self._Thread__args, **self._Thread__kwargs)
def join(self, *args, **kwargs):
super(threadWithReturn, self).join(*args, **kwargs)
return self._return
UPDATED ANSWER BELOW
This is my most popularly upvoted answer, so I decided to update with code that will run on both py2 and py3.
Additionally, I see many answers to this question that show a lack of comprehension regarding Thread.join(). Some completely fail to handle the timeout arg. But there is also a corner-case that you should be aware of regarding instances when you have (1) a target function that can return None and (2) you also pass the timeout arg to join(). Please see "TEST 4" to understand this corner case.
ThreadWithReturn class that works with py2 and py3:
import sys
from threading import Thread
from builtins import super # https://stackoverflow.com/a/30159479
_thread_target_key, _thread_args_key, _thread_kwargs_key = (
('_target', '_args', '_kwargs')
if sys.version_info >= (3, 0) else
('_Thread__target', '_Thread__args', '_Thread__kwargs')
)
class ThreadWithReturn(Thread):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self._return = None
def run(self):
target = getattr(self, _thread_target_key)
if target is not None:
self._return = target(
*getattr(self, _thread_args_key),
**getattr(self, _thread_kwargs_key)
)
def join(self, *args, **kwargs):
super().join(*args, **kwargs)
return self._return
Some sample tests are shown below:
import time, random
# TEST TARGET FUNCTION
def giveMe(arg, seconds=None):
if not seconds is None:
time.sleep(seconds)
return arg
# TEST 1
my_thread = ThreadWithReturn(target=giveMe, args=('stringy',))
my_thread.start()
returned = my_thread.join()
# (returned == 'stringy')
# TEST 2
my_thread = ThreadWithReturn(target=giveMe, args=(None,))
my_thread.start()
returned = my_thread.join()
# (returned is None)
# TEST 3
my_thread = ThreadWithReturn(target=giveMe, args=('stringy',), kwargs={'seconds': 5})
my_thread.start()
returned = my_thread.join(timeout=2)
# (returned is None) # because join() timed out before giveMe() finished
# TEST 4
my_thread = ThreadWithReturn(target=giveMe, args=(None,), kwargs={'seconds': 5})
my_thread.start()
returned = my_thread.join(timeout=random.randint(1, 10))
Can you identify the corner-case that we may possibly encounter with TEST 4?
The problem is that we expect giveMe() to return None (see TEST 2), but we also expect join() to return None if it times out.
returned is None means either:
(1) that's what giveMe() returned, or
(2) join() timed out
This example is trivial since we know that giveMe() will always return None. But in real-world instance (where the target may legitimately return None or something else) we'd want to explicitly check for what happened.
Below is how to address this corner-case:
# TEST 4
my_thread = ThreadWithReturn(target=giveMe, args=(None,), kwargs={'seconds': 5})
my_thread.start()
returned = my_thread.join(timeout=random.randint(1, 10))
if my_thread.isAlive():
# returned is None because join() timed out
# this also means that giveMe() is still running in the background
pass
# handle this based on your app's logic
else:
# join() is finished, and so is giveMe()
# BUT we could also be in a race condition, so we need to update returned, just in case
returned = my_thread.join()
Using Queue :
import threading, queue
def calc_square(num, out_queue1):
l = []
for x in num:
l.append(x*x)
out_queue1.put(l)
arr = [1,2,3,4,5,6,7,8,9,10]
out_queue1=queue.Queue()
t1=threading.Thread(target=calc_square, args=(arr,out_queue1))
t1.start()
t1.join()
print (out_queue1.get())
My solution to the problem is to wrap the function and thread in a class. Does not require using pools,queues, or c type variable passing. It is also non blocking. You check status instead. See example of how to use it at end of code.
import threading
class ThreadWorker():
'''
The basic idea is given a function create an object.
The object can then run the function in a thread.
It provides a wrapper to start it,check its status,and get data out the function.
'''
def __init__(self,func):
self.thread = None
self.data = None
self.func = self.save_data(func)
def save_data(self,func):
'''modify function to save its returned data'''
def new_func(*args, **kwargs):
self.data=func(*args, **kwargs)
return new_func
def start(self,params):
self.data = None
if self.thread is not None:
if self.thread.isAlive():
return 'running' #could raise exception here
#unless thread exists and is alive start or restart it
self.thread = threading.Thread(target=self.func,args=params)
self.thread.start()
return 'started'
def status(self):
if self.thread is None:
return 'not_started'
else:
if self.thread.isAlive():
return 'running'
else:
return 'finished'
def get_results(self):
if self.thread is None:
return 'not_started' #could return exception
else:
if self.thread.isAlive():
return 'running'
else:
return self.data
def add(x,y):
return x +y
add_worker = ThreadWorker(add)
print add_worker.start((1,2,))
print add_worker.status()
print add_worker.get_results()
Taking into consideration #iman comment on #JakeBiesinger answer I have recomposed it to have various number of threads:
from multiprocessing.pool import ThreadPool
def foo(bar, baz):
print 'hello {0}'.format(bar)
return 'foo' + baz
numOfThreads = 3
results = []
pool = ThreadPool(numOfThreads)
for i in range(0, numOfThreads):
results.append(pool.apply_async(foo, ('world', 'foo'))) # tuple of args for foo)
# do some other stuff in the main process
# ...
# ...
results = [r.get() for r in results]
print results
pool.close()
pool.join()
I'm using this wrapper, which comfortably turns any function for running in a Thread - taking care of its return value or exception. It doesn't add Queue overhead.
def threading_func(f):
"""Decorator for running a function in a thread and handling its return
value or exception"""
def start(*args, **kw):
def run():
try:
th.ret = f(*args, **kw)
except:
th.exc = sys.exc_info()
def get(timeout=None):
th.join(timeout)
if th.exc:
raise th.exc[0], th.exc[1], th.exc[2] # py2
##raise th.exc[1] #py3
return th.ret
th = threading.Thread(None, run)
th.exc = None
th.get = get
th.start()
return th
return start
Usage Examples
def f(x):
return 2.5 * x
th = threading_func(f)(4)
print("still running?:", th.is_alive())
print("result:", th.get(timeout=1.0))
#threading_func
def th_mul(a, b):
return a * b
th = th_mul("text", 2.5)
try:
print(th.get())
except TypeError:
print("exception thrown ok.")
Notes on threading module
Comfortable return value & exception handling of a threaded function is a frequent "Pythonic" need and should indeed already be offered by the threading module - possibly directly in the standard Thread class. ThreadPool has way too much overhead for simple tasks - 3 managing threads, lots of bureaucracy. Unfortunately Thread's layout was copied from Java originally - which you see e.g. from the still useless 1st (!) constructor parameter group.
Based of what kindall mentioned, here's the more generic solution that works with Python3.
import threading
class ThreadWithReturnValue(threading.Thread):
def __init__(self, *init_args, **init_kwargs):
threading.Thread.__init__(self, *init_args, **init_kwargs)
self._return = None
def run(self):
self._return = self._target(*self._args, **self._kwargs)
def join(self):
threading.Thread.join(self)
return self._return
Usage
th = ThreadWithReturnValue(target=requests.get, args=('http://www.google.com',))
th.start()
response = th.join()
response.status_code # => 200
join always return None, i think you should subclass Thread to handle return codes and so.
You can define a mutable above the scope of the threaded function, and add the result to that. (I also modified the code to be python3 compatible)
returns = {}
def foo(bar):
print('hello {0}'.format(bar))
returns[bar] = 'foo'
from threading import Thread
t = Thread(target=foo, args=('world!',))
t.start()
t.join()
print(returns)
This returns {'world!': 'foo'}
If you use the function input as the key to your results dict, every unique input is guaranteed to give an entry in the results
Define your target to
1) take an argument q
2) replace any statements return foo with q.put(foo); return
so a function
def func(a):
ans = a * a
return ans
would become
def func(a, q):
ans = a * a
q.put(ans)
return
and then you would proceed as such
from Queue import Queue
from threading import Thread
ans_q = Queue()
arg_tups = [(i, ans_q) for i in xrange(10)]
threads = [Thread(target=func, args=arg_tup) for arg_tup in arg_tups]
_ = [t.start() for t in threads]
_ = [t.join() for t in threads]
results = [q.get() for _ in xrange(len(threads))]
And you can use function decorators/wrappers to make it so you can use your existing functions as target without modifying them, but follow this basic scheme.
GuySoft's idea is great, but I think the object does not necessarily have to inherit from Thread and start() could be removed from interface:
from threading import Thread
import queue
class ThreadWithReturnValue(object):
def __init__(self, target=None, args=(), **kwargs):
self._que = queue.Queue()
self._t = Thread(target=lambda q,arg1,kwargs1: q.put(target(*arg1, **kwargs1)) ,
args=(self._que, args, kwargs), )
self._t.start()
def join(self):
self._t.join()
return self._que.get()
def foo(bar):
print('hello {0}'.format(bar))
return "foo"
twrv = ThreadWithReturnValue(target=foo, args=('world!',))
print(twrv.join()) # prints foo
This is a pretty old question, but I wanted to share a simple solution that has worked for me and helped my dev process.
The methodology behind this answer is the fact that the "new" target function, inner is assigning the result of the original function (passed through the __init__ function) to the result instance attribute of the wrapper through something called closure.
This allows the wrapper class to hold onto the return value for callers to access at anytime.
NOTE: This method doesn't need to use any mangled methods or private methods of the threading.Thread class, although yield functions have not been considered (OP did not mention yield functions).
Enjoy!
from threading import Thread as _Thread
class ThreadWrapper:
def __init__(self, target, *args, **kwargs):
self.result = None
self._target = self._build_threaded_fn(target)
self.thread = _Thread(
target=self._target,
*args,
**kwargs
)
def _build_threaded_fn(self, func):
def inner(*args, **kwargs):
self.result = func(*args, **kwargs)
return inner
Additionally, you can run pytest (assuming you have it installed) with the following code to demonstrate the results:
import time
from commons import ThreadWrapper
def test():
def target():
time.sleep(1)
return 'Hello'
wrapper = ThreadWrapper(target=target)
wrapper.thread.start()
r = wrapper.result
assert r is None
time.sleep(2)
r = wrapper.result
assert r == 'Hello'
As mentioned multiprocessing pool is much slower than basic threading. Using queues as proposeded in some answers here is a very effective alternative. I have use it with dictionaries in order to be able run a lot of small threads and recuperate multiple answers by combining them with dictionaries:
#!/usr/bin/env python3
import threading
# use Queue for python2
import queue
import random
LETTERS = 'abcdefghijklmnopqrstuvwxyz'
LETTERS = [ x for x in LETTERS ]
NUMBERS = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
def randoms(k, q):
result = dict()
result['letter'] = random.choice(LETTERS)
result['number'] = random.choice(NUMBERS)
q.put({k: result})
threads = list()
q = queue.Queue()
results = dict()
for name in ('alpha', 'oscar', 'yankee',):
threads.append( threading.Thread(target=randoms, args=(name, q)) )
threads[-1].start()
_ = [ t.join() for t in threads ]
while not q.empty():
results.update(q.get())
print(results)
Here is the version that I created of #Kindall's answer.
This version makes it so that all you have to do is input your command with arguments to create the new thread.
This was made with Python 3.8:
from threading import Thread
from typing import Any
def test(plug, plug2, plug3):
print(f"hello {plug}")
print(f'I am the second plug : {plug2}')
print(plug3)
return 'I am the return Value!'
def test2(msg):
return f'I am from the second test: {msg}'
def test3():
print('hello world')
def NewThread(com, Returning: bool, *arguments) -> Any:
"""
Will create a new thread for a function/command.
:param com: Command to be Executed
:param arguments: Arguments to be sent to Command
:param Returning: True/False Will this command need to return anything
"""
class NewThreadWorker(Thread):
def __init__(self, group = None, target = None, name = None, args = (), kwargs = None, *,
daemon = None):
Thread.__init__(self, group, target, name, args, kwargs, daemon = daemon)
self._return = None
def run(self):
if self._target is not None:
self._return = self._target(*self._args, **self._kwargs)
def join(self):
Thread.join(self)
return self._return
ntw = NewThreadWorker(target = com, args = (*arguments,))
ntw.start()
if Returning:
return ntw.join()
if __name__ == "__main__":
print(NewThread(test, True, 'hi', 'test', test2('hi')))
NewThread(test3, True)
You can use pool.apply_async() of ThreadPool() to return the value from test() as shown below:
from multiprocessing.pool import ThreadPool
def test(num1, num2):
return num1 + num2
pool = ThreadPool(processes=1) # Here
result = pool.apply_async(test, (2, 3)) # Here
print(result.get()) # 5
And, you can also use submit() of concurrent.futures.ThreadPoolExecutor() to return the value from test() as shown below:
from concurrent.futures import ThreadPoolExecutor
def test(num1, num2):
return num1 + num2
with ThreadPoolExecutor(max_workers=1) as executor:
future = executor.submit(test, 2, 3) # Here
print(future.result()) # 5
And, instead of return, you can use the array result as shown below:
from threading import Thread
def test(num1, num2, r):
r[0] = num1 + num2 # Instead of "return"
result = [None] # Here
thread = Thread(target=test, args=(2, 3, result))
thread.start()
thread.join()
print(result[0]) # 5
And instead of return, you can also use the queue result as shown below:
from threading import Thread
import queue
def test(num1, num2, q):
q.put(num1 + num2) # Instead of "return"
queue = queue.Queue() # Here
thread = Thread(target=test, args=(2, 3, queue))
thread.start()
thread.join()
print(queue.get()) # '5'
The shortest and simplest way I've found to do this is to take advantage of Python classes and their dynamic properties. You can retrieve the current thread from within the context of your spawned thread using threading.current_thread(), and assign the return value to a property.
import threading
def some_target_function():
# Your code here.
threading.current_thread().return_value = "Some return value."
your_thread = threading.Thread(target=some_target_function)
your_thread.start()
your_thread.join()
return_value = your_thread.return_value
print(return_value)
One usual solution is to wrap your function foo with a decorator like
result = queue.Queue()
def task_wrapper(*args):
result.put(target(*args))
Then the whole code may looks like that
result = queue.Queue()
def task_wrapper(*args):
result.put(target(*args))
threads = [threading.Thread(target=task_wrapper, args=args) for args in args_list]
for t in threads:
t.start()
while(True):
if(len(threading.enumerate()) < max_num):
break
for t in threads:
t.join()
return result
Note
One important issue is that the return values may be unorderred.
(In fact, the return value is not necessarily saved to the queue, since you can choose arbitrary thread-safe data structure )
Kindall's answer in Python3
class ThreadWithReturnValue(Thread):
def __init__(self, group=None, target=None, name=None,
args=(), kwargs={}, *, daemon=None):
Thread.__init__(self, group, target, name, args, kwargs, daemon)
self._return = None
def run(self):
try:
if self._target:
self._return = self._target(*self._args, **self._kwargs)
finally:
del self._target, self._args, self._kwargs
def join(self,timeout=None):
Thread.join(self,timeout)
return self._return
I know this thread is old.... but I faced the same problem... If you are willing to use thread.join()
import threading
class test:
def __init__(self):
self.msg=""
def hello(self,bar):
print('hello {}'.format(bar))
self.msg="foo"
def main(self):
thread = threading.Thread(target=self.hello, args=('world!',))
thread.start()
thread.join()
print(self.msg)
g=test()
g.main()
Best way... Define a global variable, then change the variable in the threaded function. Nothing to pass in or retrieve back
from threading import Thread
# global var
radom_global_var = 5
def function():
global random_global_var
random_global_var += 1
domath = Thread(target=function)
domath.start()
domath.join()
print(random_global_var)
# result: 6
I researched first and couldn't find an answer to my question. I am trying to run multiple functions in parallel in Python.
I have something like this:
files.py
import common #common is a util class that handles all the IO stuff
dir1 = 'C:\folder1'
dir2 = 'C:\folder2'
filename = 'test.txt'
addFiles = [25, 5, 15, 35, 45, 25, 5, 15, 35, 45]
def func1():
c = common.Common()
for i in range(len(addFiles)):
c.createFiles(addFiles[i], filename, dir1)
c.getFiles(dir1)
time.sleep(10)
c.removeFiles(addFiles[i], dir1)
c.getFiles(dir1)
def func2():
c = common.Common()
for i in range(len(addFiles)):
c.createFiles(addFiles[i], filename, dir2)
c.getFiles(dir2)
time.sleep(10)
c.removeFiles(addFiles[i], dir2)
c.getFiles(dir2)
I want to call func1 and func2 and have them run at the same time. The functions do not interact with each other or on the same object. Right now I have to wait for func1 to finish before func2 to start. How do I do something like below:
process.py
from files import func1, func2
runBothFunc(func1(), func2())
I want to be able to create both directories pretty close to the same time because every min I am counting how many files are being created. If the directory isn't there it will throw off my timing.
You could use threading or multiprocessing.
Due to peculiarities of CPython, threading is unlikely to achieve true parallelism. For this reason, multiprocessing is generally a better bet.
Here is a complete example:
from multiprocessing import Process
def func1():
print 'func1: starting'
for i in xrange(10000000): pass
print 'func1: finishing'
def func2():
print 'func2: starting'
for i in xrange(10000000): pass
print 'func2: finishing'
if __name__ == '__main__':
p1 = Process(target=func1)
p1.start()
p2 = Process(target=func2)
p2.start()
p1.join()
p2.join()
The mechanics of starting/joining child processes can easily be encapsulated into a function along the lines of your runBothFunc:
def runInParallel(*fns):
proc = []
for fn in fns:
p = Process(target=fn)
p.start()
proc.append(p)
for p in proc:
p.join()
runInParallel(func1, func2)
If your functions are mainly doing I/O work (and less CPU work) and you have Python 3.2+, you can use a ThreadPoolExecutor:
from concurrent.futures import ThreadPoolExecutor
def run_io_tasks_in_parallel(tasks):
with ThreadPoolExecutor() as executor:
running_tasks = [executor.submit(task) for task in tasks]
for running_task in running_tasks:
running_task.result()
run_io_tasks_in_parallel([
lambda: print('IO task 1 running!'),
lambda: print('IO task 2 running!'),
])
If your functions are mainly doing CPU work (and less I/O work) and you have Python 2.6+, you can use the multiprocessing module:
from multiprocessing import Process
def run_cpu_tasks_in_parallel(tasks):
running_tasks = [Process(target=task) for task in tasks]
for running_task in running_tasks:
running_task.start()
for running_task in running_tasks:
running_task.join()
run_cpu_tasks_in_parallel([
lambda: print('CPU task 1 running!'),
lambda: print('CPU task 2 running!'),
])
This can be done elegantly with Ray, a system that allows you to easily parallelize and distribute your Python code.
To parallelize your example, you'd need to define your functions with the #ray.remote decorator, and then invoke them with .remote.
import ray
ray.init()
dir1 = 'C:\\folder1'
dir2 = 'C:\\folder2'
filename = 'test.txt'
addFiles = [25, 5, 15, 35, 45, 25, 5, 15, 35, 45]
# Define the functions.
# You need to pass every global variable used by the function as an argument.
# This is needed because each remote function runs in a different process,
# and thus it does not have access to the global variables defined in
# the current process.
#ray.remote
def func1(filename, addFiles, dir):
# func1() code here...
#ray.remote
def func2(filename, addFiles, dir):
# func2() code here...
# Start two tasks in the background and wait for them to finish.
ray.get([func1.remote(filename, addFiles, dir1), func2.remote(filename, addFiles, dir2)])
If you pass the same argument to both functions and the argument is large, a more efficient way to do this is using ray.put(). This avoids the large argument to be serialized twice and to create two memory copies of it:
largeData_id = ray.put(largeData)
ray.get([func1(largeData_id), func2(largeData_id)])
Important - If func1() and func2() return results, you need to rewrite the code as follows:
ret_id1 = func1.remote(filename, addFiles, dir1)
ret_id2 = func2.remote(filename, addFiles, dir2)
ret1, ret2 = ray.get([ret_id1, ret_id2])
There are a number of advantages of using Ray over the multiprocessing module. In particular, the same code will run on a single machine as well as on a cluster of machines. For more advantages of Ray see this related post.
Seems like you have a single function that you need to call on two different parameters. This can be elegantly done using a combination of concurrent.futures and map with Python 3.2+
import time
from concurrent.futures import ThreadPoolExecutor, ProcessPoolExecutor
def sleep_secs(seconds):
time.sleep(seconds)
print(f'{seconds} has been processed')
secs_list = [2,4, 6, 8, 10, 12]
Now, if your operation is IO bound, then you can use the ThreadPoolExecutor as such:
with ThreadPoolExecutor() as executor:
results = executor.map(sleep_secs, secs_list)
Note how map is used here to map your function to the list of arguments.
Now, If your function is CPU bound, then you can use ProcessPoolExecutor
with ProcessPoolExecutor() as executor:
results = executor.map(sleep_secs, secs_list)
If you are not sure, you can simply try both and see which one gives you better results.
Finally, if you are looking to print out your results, you can simply do this:
with ThreadPoolExecutor() as executor:
results = executor.map(sleep_secs, secs_list)
for result in results:
print(result)
In 2021 the easiest way is to use asyncio:
import asyncio, time
async def say_after(delay, what):
await asyncio.sleep(delay)
print(what)
async def main():
task1 = asyncio.create_task(
say_after(4, 'hello'))
task2 = asyncio.create_task(
say_after(3, 'world'))
print(f"started at {time.strftime('%X')}")
# Wait until both tasks are completed (should take
# around 2 seconds.)
await task1
await task2
print(f"finished at {time.strftime('%X')}")
asyncio.run(main())
References:
[1] https://docs.python.org/3/library/asyncio-task.html
If you are a windows user and using python 3, then this post will help you to do parallel programming in python.when you run a usual multiprocessing library's pool programming, you will get an error regarding the main function in your program. This is because the fact that windows has no fork() functionality. The below post is giving a solution to the mentioned problem .
http://python.6.x6.nabble.com/Multiprocessing-Pool-woes-td5047050.html
Since I was using the python 3, I changed the program a little like this:
from types import FunctionType
import marshal
def _applicable(*args, **kwargs):
name = kwargs['__pw_name']
code = marshal.loads(kwargs['__pw_code'])
gbls = globals() #gbls = marshal.loads(kwargs['__pw_gbls'])
defs = marshal.loads(kwargs['__pw_defs'])
clsr = marshal.loads(kwargs['__pw_clsr'])
fdct = marshal.loads(kwargs['__pw_fdct'])
func = FunctionType(code, gbls, name, defs, clsr)
func.fdct = fdct
del kwargs['__pw_name']
del kwargs['__pw_code']
del kwargs['__pw_defs']
del kwargs['__pw_clsr']
del kwargs['__pw_fdct']
return func(*args, **kwargs)
def make_applicable(f, *args, **kwargs):
if not isinstance(f, FunctionType): raise ValueError('argument must be a function')
kwargs['__pw_name'] = f.__name__ # edited
kwargs['__pw_code'] = marshal.dumps(f.__code__) # edited
kwargs['__pw_defs'] = marshal.dumps(f.__defaults__) # edited
kwargs['__pw_clsr'] = marshal.dumps(f.__closure__) # edited
kwargs['__pw_fdct'] = marshal.dumps(f.__dict__) # edited
return _applicable, args, kwargs
def _mappable(x):
x,name,code,defs,clsr,fdct = x
code = marshal.loads(code)
gbls = globals() #gbls = marshal.loads(gbls)
defs = marshal.loads(defs)
clsr = marshal.loads(clsr)
fdct = marshal.loads(fdct)
func = FunctionType(code, gbls, name, defs, clsr)
func.fdct = fdct
return func(x)
def make_mappable(f, iterable):
if not isinstance(f, FunctionType): raise ValueError('argument must be a function')
name = f.__name__ # edited
code = marshal.dumps(f.__code__) # edited
defs = marshal.dumps(f.__defaults__) # edited
clsr = marshal.dumps(f.__closure__) # edited
fdct = marshal.dumps(f.__dict__) # edited
return _mappable, ((i,name,code,defs,clsr,fdct) for i in iterable)
After this function , the above problem code is also changed a little like this:
from multiprocessing import Pool
from poolable import make_applicable, make_mappable
def cube(x):
return x**3
if __name__ == "__main__":
pool = Pool(processes=2)
results = [pool.apply_async(*make_applicable(cube,x)) for x in range(1,7)]
print([result.get(timeout=10) for result in results])
And I got the output as :
[1, 8, 27, 64, 125, 216]
I am thinking that this post may be useful for some of the windows users.
There's no way to guarantee that two functions will execute in sync with each other which seems to be what you want to do.
The best you can do is to split up the function into several steps, then wait for both to finish at critical synchronization points using Process.join like #aix's answer mentions.
This is better than time.sleep(10) because you can't guarantee exact timings. With explicitly waiting, you're saying that the functions must be done executing that step before moving to the next, instead of assuming it will be done within 10ms which isn't guaranteed based on what else is going on on the machine.
(about How can I simultaneously run two (or more) functions in python?)
With asyncio, sync/async tasks could be run concurrently by:
import asyncio
import time
def function1():
# performing blocking tasks
while True:
print("function 1: blocking task ...")
time.sleep(1)
async def function2():
# perform non-blocking tasks
while True:
print("function 2: non-blocking task ...")
await asyncio.sleep(1)
async def main():
loop = asyncio.get_running_loop()
await asyncio.gather(
# https://docs.python.org/3/library/asyncio-eventloop.html#asyncio.loop.run_in_executor
loop.run_in_executor(None, function1),
function2(),
)
if __name__ == '__main__':
asyncio.run(main())
How can I get the following to work? The main point is that I want to run a method (and not a function) asynchronously.
from multiprocessing import Pool
class Async:
def __init__(self, pool):
self.pool = pool
self.run()
def run(self):
p.apply_async(self.f, (10, ))
def f(self, x):
print x*x
if __name__ == '__main__':
p = Pool(5)
a = Async(p)
p.close()
p.join()
This prints nothing.
The problem appears to be due to the fact that multiprocessing needs to pickle self.f while bound methods are not picklable. There is a discussion on how to solve the problem here.
The apply_async apparently creates an exception which is put inside the future returned. That's why nothing is printed. If a get is executed on the future, then the exception is raised.
Its definitely possible to thread class methods using a threadpool in python 2 - the following programme did what I would expect.
#!/usr/bin/env python
from multiprocessing.pool import ThreadPool
class TestAsync():
def __init__(self):
pool = ThreadPool(processes = 2)
async_completions = []
for a in range(2):
async_completions.append(pool.apply_async(self.print_int, ( a,)))
for completion in async_completions:
res = completion.get()
print("res = %d" % res)
def print_int(self, value):
print(value)
return (value*10)
a = TestAsync()
I am trying to do some computations using the multiprocessing module in python 2.7.2.
My code is like this:
from multiprocessing import Pool
import sys
sys.setrecursionlimit(10000)
partitions = []
class Partitions:
parts = {} #My goal is to use this dict to speed
#up calculations in every process that
#uses it, without having to build it up
#from nothing each time
def __init__(self):
pass
def p1(self, k, n):
if (k,n) in Partitions.parts:
return Partitions.parts[(k, n)]
if k>n:
return 0
if k==n:
return 1
Partitions.parts[(k,n)] = self.p1(k+1, n) + self.p1(k, n-k)
return Partitions.parts[(k,n)]
def P(self, n):
result = 0
for k in xrange(1,n/2 + 1):
result += self.p1(k, n-k)
return 1 + result
p = Partitions()
def log(results):
if results:
partitions.extend(results)
return None
def partWorker(start,stop):
ps = []
for n in xrange(start, stop):
ps.append(((1,n), p.P(n)))
return ps
def main():
pool = Pool()
step = 150
for i in xrange(0,301,step):
pool.apply_async(partWorker, (i, i+step), callback = log)
pool.close()
pool.join()
return None
if __name__=="__main__":
main()
I am new to this, I basically copied the format of the prime code on this page:
python prime crunching: processing pool is slower?
Can I get process running in each core all looking at the same dictionary to assist their
calculations? The way it behaves now, each process creates it's own dictionaries and it eats up ram like crazy.
I'm not sure if this is what you want ... but, take a look at multiprocessing.Manager ( http://docs.python.org/library/multiprocessing.html#sharing-state-between-processes ). Managers allow you to share a dict between processes.
When I run something like:
from multiprocessing import Pool
p = Pool(5)
def f(x):
return x*x
p.map(f, [1,2,3])
it works fine. However, putting this as a function of a class:
class calculate(object):
def run(self):
def f(x):
return x*x
p = Pool()
return p.map(f, [1,2,3])
cl = calculate()
print cl.run()
Gives me the following error:
Exception in thread Thread-1:
Traceback (most recent call last):
File "/sw/lib/python2.6/threading.py", line 532, in __bootstrap_inner
self.run()
File "/sw/lib/python2.6/threading.py", line 484, in run
self.__target(*self.__args, **self.__kwargs)
File "/sw/lib/python2.6/multiprocessing/pool.py", line 225, in _handle_tasks
put(task)
PicklingError: Can't pickle <type 'function'>: attribute lookup __builtin__.function failed
I've seen a post from Alex Martelli dealing with the same kind of problem, but it wasn't explicit enough.
I could not use the code posted so far because code using "multiprocessing.Pool" do not work with lambda expressions and code not using "multiprocessing.Pool" spawn as many processes as there are work items.
I adapted the code s.t. it spawns a predefined amount of workers and only iterates through the input list if there exists an idle worker. I also enabled the "daemon" mode for the workers s.t. ctrl-c works as expected.
import multiprocessing
def fun(f, q_in, q_out):
while True:
i, x = q_in.get()
if i is None:
break
q_out.put((i, f(x)))
def parmap(f, X, nprocs=multiprocessing.cpu_count()):
q_in = multiprocessing.Queue(1)
q_out = multiprocessing.Queue()
proc = [multiprocessing.Process(target=fun, args=(f, q_in, q_out))
for _ in range(nprocs)]
for p in proc:
p.daemon = True
p.start()
sent = [q_in.put((i, x)) for i, x in enumerate(X)]
[q_in.put((None, None)) for _ in range(nprocs)]
res = [q_out.get() for _ in range(len(sent))]
[p.join() for p in proc]
return [x for i, x in sorted(res)]
if __name__ == '__main__':
print(parmap(lambda i: i * 2, [1, 2, 3, 4, 6, 7, 8]))
Multiprocessing and pickling is broken and limited unless you jump outside the standard library.
If you use a fork of multiprocessing called pathos.multiprocesssing, you can directly use classes and class methods in multiprocessing's map functions. This is because dill is used instead of pickle or cPickle, and dill can serialize almost anything in python.
pathos.multiprocessing also provides an asynchronous map function… and it can map functions with multiple arguments (e.g. map(math.pow, [1,2,3], [4,5,6]))
See discussions:
What can multiprocessing and dill do together?
and:
http://matthewrocklin.com/blog/work/2013/12/05/Parallelism-and-Serialization
It even handles the code you wrote initially, without modification, and from the interpreter. Why do anything else that's more fragile and specific to a single case?
>>> from pathos.multiprocessing import ProcessingPool as Pool
>>> class calculate(object):
... def run(self):
... def f(x):
... return x*x
... p = Pool()
... return p.map(f, [1,2,3])
...
>>> cl = calculate()
>>> print cl.run()
[1, 4, 9]
Get the code here:
https://github.com/uqfoundation/pathos
And, just to show off a little more of what it can do:
>>> from pathos.multiprocessing import ProcessingPool as Pool
>>>
>>> p = Pool(4)
>>>
>>> def add(x,y):
... return x+y
...
>>> x = [0,1,2,3]
>>> y = [4,5,6,7]
>>>
>>> p.map(add, x, y)
[4, 6, 8, 10]
>>>
>>> class Test(object):
... def plus(self, x, y):
... return x+y
...
>>> t = Test()
>>>
>>> p.map(Test.plus, [t]*4, x, y)
[4, 6, 8, 10]
>>>
>>> res = p.amap(t.plus, x, y)
>>> res.get()
[4, 6, 8, 10]
I also was annoyed by restrictions on what sort of functions pool.map could accept. I wrote the following to circumvent this. It appears to work, even for recursive use of parmap.
from multiprocessing import Process, Pipe
from itertools import izip
def spawn(f):
def fun(pipe, x):
pipe.send(f(x))
pipe.close()
return fun
def parmap(f, X):
pipe = [Pipe() for x in X]
proc = [Process(target=spawn(f), args=(c, x)) for x, (p, c) in izip(X, pipe)]
[p.start() for p in proc]
[p.join() for p in proc]
return [p.recv() for (p, c) in pipe]
if __name__ == '__main__':
print parmap(lambda x: x**x, range(1, 5))
There is currently no solution to your problem, as far as I know: the function that you give to map() must be accessible through an import of your module. This is why robert's code works: the function f() can be obtained by importing the following code:
def f(x):
return x*x
class Calculate(object):
def run(self):
p = Pool()
return p.map(f, [1,2,3])
if __name__ == '__main__':
cl = Calculate()
print cl.run()
I actually added a "main" section, because this follows the recommendations for the Windows platform ("Make sure that the main module can be safely imported by a new Python interpreter without causing unintended side effects").
I also added an uppercase letter in front of Calculate, so as to follow PEP 8. :)
The solution by mrule is correct but has a bug: if the child sends back a large amount of data, it can fill the pipe's buffer, blocking on the child's pipe.send(), while the parent is waiting for the child to exit on pipe.join(). The solution is to read the child's data before join()ing the child. Furthermore the child should close the parent's end of the pipe to prevent a deadlock. The code below fixes that. Also be aware that this parmap creates one process per element in X. A more advanced solution is to use multiprocessing.cpu_count() to divide X into a number of chunks, and then merge the results before returning. I leave that as an exercise to the reader so as not to spoil the conciseness of the nice answer by mrule. ;)
from multiprocessing import Process, Pipe
from itertools import izip
def spawn(f):
def fun(ppipe, cpipe,x):
ppipe.close()
cpipe.send(f(x))
cpipe.close()
return fun
def parmap(f,X):
pipe=[Pipe() for x in X]
proc=[Process(target=spawn(f),args=(p,c,x)) for x,(p,c) in izip(X,pipe)]
[p.start() for p in proc]
ret = [p.recv() for (p,c) in pipe]
[p.join() for p in proc]
return ret
if __name__ == '__main__':
print parmap(lambda x:x**x,range(1,5))
I've also struggled with this. I had functions as data members of a class, as a simplified example:
from multiprocessing import Pool
import itertools
pool = Pool()
class Example(object):
def __init__(self, my_add):
self.f = my_add
def add_lists(self, list1, list2):
# Needed to do something like this (the following line won't work)
return pool.map(self.f,list1,list2)
I needed to use the function self.f in a Pool.map() call from within the same class and self.f did not take a tuple as an argument. Since this function was embedded in a class, it was not clear to me how to write the type of wrapper other answers suggested.
I solved this problem by using a different wrapper that takes a tuple/list, where the first element is the function, and the remaining elements are the arguments to that function, called eval_func_tuple(f_args). Using this, the problematic line can be replaced by return pool.map(eval_func_tuple, itertools.izip(itertools.repeat(self.f), list1, list2)). Here is the full code:
File: util.py
def add(a, b): return a+b
def eval_func_tuple(f_args):
"""Takes a tuple of a function and args, evaluates and returns result"""
return f_args[0](*f_args[1:])
File: main.py
from multiprocessing import Pool
import itertools
import util
pool = Pool()
class Example(object):
def __init__(self, my_add):
self.f = my_add
def add_lists(self, list1, list2):
# The following line will now work
return pool.map(util.eval_func_tuple,
itertools.izip(itertools.repeat(self.f), list1, list2))
if __name__ == '__main__':
myExample = Example(util.add)
list1 = [1, 2, 3]
list2 = [10, 20, 30]
print myExample.add_lists(list1, list2)
Running main.py will give [11, 22, 33]. Feel free to improve this, for example eval_func_tuple could also be modified to take keyword arguments.
On another note, in another answers, the function "parmap" can be made more efficient for the case of more Processes than number of CPUs available. I'm copying an edited version below. This is my first post and I wasn't sure if I should directly edit the original answer. I also renamed some variables.
from multiprocessing import Process, Pipe
from itertools import izip
def spawn(f):
def fun(pipe,x):
pipe.send(f(x))
pipe.close()
return fun
def parmap(f,X):
pipe=[Pipe() for x in X]
processes=[Process(target=spawn(f),args=(c,x)) for x,(p,c) in izip(X,pipe)]
numProcesses = len(processes)
processNum = 0
outputList = []
while processNum < numProcesses:
endProcessNum = min(processNum+multiprocessing.cpu_count(), numProcesses)
for proc in processes[processNum:endProcessNum]:
proc.start()
for proc in processes[processNum:endProcessNum]:
proc.join()
for proc,c in pipe[processNum:endProcessNum]:
outputList.append(proc.recv())
processNum = endProcessNum
return outputList
if __name__ == '__main__':
print parmap(lambda x:x**x,range(1,5))
I know that this question was asked 8 years and 10 months ago but I want to present you my solution:
from multiprocessing import Pool
class Test:
def __init__(self):
self.main()
#staticmethod
def methodForMultiprocessing(x):
print(x*x)
def main(self):
if __name__ == "__main__":
p = Pool()
p.map(Test.methodForMultiprocessing, list(range(1, 11)))
p.close()
TestObject = Test()
You just need to make your class function into a static method. But it's also possible with a class method:
from multiprocessing import Pool
class Test:
def __init__(self):
self.main()
#classmethod
def methodForMultiprocessing(cls, x):
print(x*x)
def main(self):
if __name__ == "__main__":
p = Pool()
p.map(Test.methodForMultiprocessing, list(range(1, 11)))
p.close()
TestObject = Test()
Tested in Python 3.7.3
I know this was asked over 6 years ago now, but just wanted to add my solution, as some of the suggestions above seem horribly complicated, but my solution was actually very simple.
All I had to do was wrap the pool.map() call to a helper function. Passing the class object along with args for the method as a tuple, which looked a bit like this.
def run_in_parallel(args):
return args[0].method(args[1])
myclass = MyClass()
method_args = [1,2,3,4,5,6]
args_map = [ (myclass, arg) for arg in method_args ]
pool = Pool()
pool.map(run_in_parallel, args_map)
I took klaus se's and aganders3's answer, and made a documented module that is more readable and holds in one file. You can just add it to your project. It even has an optional progress bar !
"""
The ``processes`` module provides some convenience functions
for using parallel processes in python.
Adapted from http://stackoverflow.com/a/16071616/287297
Example usage:
print prll_map(lambda i: i * 2, [1, 2, 3, 4, 6, 7, 8], 32, verbose=True)
Comments:
"It spawns a predefined amount of workers and only iterates through the input list
if there exists an idle worker. I also enabled the "daemon" mode for the workers so
that KeyboardInterupt works as expected."
Pitfalls: all the stdouts are sent back to the parent stdout, intertwined.
Alternatively, use this fork of multiprocessing:
https://github.com/uqfoundation/multiprocess
"""
# Modules #
import multiprocessing
from tqdm import tqdm
################################################################################
def apply_function(func_to_apply, queue_in, queue_out):
while not queue_in.empty():
num, obj = queue_in.get()
queue_out.put((num, func_to_apply(obj)))
################################################################################
def prll_map(func_to_apply, items, cpus=None, verbose=False):
# Number of processes to use #
if cpus is None: cpus = min(multiprocessing.cpu_count(), 32)
# Create queues #
q_in = multiprocessing.Queue()
q_out = multiprocessing.Queue()
# Process list #
new_proc = lambda t,a: multiprocessing.Process(target=t, args=a)
processes = [new_proc(apply_function, (func_to_apply, q_in, q_out)) for x in range(cpus)]
# Put all the items (objects) in the queue #
sent = [q_in.put((i, x)) for i, x in enumerate(items)]
# Start them all #
for proc in processes:
proc.daemon = True
proc.start()
# Display progress bar or not #
if verbose:
results = [q_out.get() for x in tqdm(range(len(sent)))]
else:
results = [q_out.get() for x in range(len(sent))]
# Wait for them to finish #
for proc in processes: proc.join()
# Return results #
return [x for i, x in sorted(results)]
################################################################################
def test():
def slow_square(x):
import time
time.sleep(2)
return x**2
objs = range(20)
squares = prll_map(slow_square, objs, 4, verbose=True)
print "Result: %s" % squares
EDIT: Added #alexander-mcfarlane suggestion and a test function
Functions defined in classes (even within functions within classes) don't really pickle. However, this works:
def f(x):
return x*x
class calculate(object):
def run(self):
p = Pool()
return p.map(f, [1,2,3])
cl = calculate()
print cl.run()
I modified klaus se's method because while it was working for me with small lists, it would hang when the number of items was ~1000 or greater. Instead of pushing the jobs one at a time with the None stop condition, I load up the input queue all at once and just let the processes munch on it until it's empty.
from multiprocessing import cpu_count, Queue, Process
def apply_func(f, q_in, q_out):
while not q_in.empty():
i, x = q_in.get()
q_out.put((i, f(x)))
# map a function using a pool of processes
def parmap(f, X, nprocs = cpu_count()):
q_in, q_out = Queue(), Queue()
proc = [Process(target=apply_func, args=(f, q_in, q_out)) for _ in range(nprocs)]
sent = [q_in.put((i, x)) for i, x in enumerate(X)]
[p.start() for p in proc]
res = [q_out.get() for _ in sent]
[p.join() for p in proc]
return [x for i,x in sorted(res)]
Edit: unfortunately now I am running into this error on my system: Multiprocessing Queue maxsize limit is 32767, hopefully the workarounds there will help.
You can run your code without any issues if you somehow manually ignore the Pool object from the list of objects in the class because it is not pickleable as the error says. You can do this with the __getstate__ function (look here too) as follow. The Pool object will try to find the __getstate__ and __setstate__ functions and execute them if it finds it when you run map, map_async etc:
class calculate(object):
def __init__(self):
self.p = Pool()
def __getstate__(self):
self_dict = self.__dict__.copy()
del self_dict['p']
return self_dict
def __setstate__(self, state):
self.__dict__.update(state)
def f(self, x):
return x*x
def run(self):
return self.p.map(self.f, [1,2,3])
Then do:
cl = calculate()
cl.run()
will give you the output:
[1, 4, 9]
I've tested the above code in Python 3.x and it works.
Here is my solution, which I think is a bit less hackish than most others here. It is similar to nightowl's answer.
someclasses = [MyClass(), MyClass(), MyClass()]
def method_caller(some_object, some_method='the method'):
return getattr(some_object, some_method)()
othermethod = partial(method_caller, some_method='othermethod')
with Pool(6) as pool:
result = pool.map(othermethod, someclasses)
This may not be a very good solution but in my case, I solve it like this.
from multiprocessing import Pool
def foo1(data):
self = data.get('slf')
lst = data.get('lst')
return sum(lst) + self.foo2()
class Foo(object):
def __init__(self, a, b):
self.a = a
self.b = b
def foo2(self):
return self.a**self.b
def foo(self):
p = Pool(5)
lst = [1, 2, 3]
result = p.map(foo1, (dict(slf=self, lst=lst),))
return result
if __name__ == '__main__':
print(Foo(2, 4).foo())
I had to pass self to my function as I have to access attributes and functions of my class through that function. This is working for me. Corrections and suggestions are always welcome.
Here is a boilerplate I wrote for using multiprocessing Pool in python3, specifically python3.7.7 was used to run the tests. I got my fastest runs using imap_unordered. Just plug in your scenario and try it out. You can use timeit or just time.time() to figure out which works best for you.
import multiprocessing
import time
NUMBER_OF_PROCESSES = multiprocessing.cpu_count()
MP_FUNCTION = 'starmap' # 'imap_unordered' or 'starmap' or 'apply_async'
def process_chunk(a_chunk):
print(f"processig mp chunk {a_chunk}")
return a_chunk
map_jobs = [1, 2, 3, 4]
result_sum = 0
s = time.time()
if MP_FUNCTION == 'imap_unordered':
pool = multiprocessing.Pool(processes=NUMBER_OF_PROCESSES)
for i in pool.imap_unordered(process_chunk, map_jobs):
result_sum += i
elif MP_FUNCTION == 'starmap':
pool = multiprocessing.Pool(processes=NUMBER_OF_PROCESSES)
try:
map_jobs = [(i, ) for i in map_jobs]
result_sum = pool.starmap(process_chunk, map_jobs)
result_sum = sum(result_sum)
finally:
pool.close()
pool.join()
elif MP_FUNCTION == 'apply_async':
with multiprocessing.Pool(processes=NUMBER_OF_PROCESSES) as pool:
result_sum = [pool.apply_async(process_chunk, [i, ]).get() for i in map_jobs]
result_sum = sum(result_sum)
print(f"result_sum is {result_sum}, took {time.time() - s}s")
In the above scenario imap_unordered actually seems to perform the worst for me. Try out your case and benchmark it on the machine you plan to run it on. Also read up on Process Pools. Cheers!
I'm not sure if this approach has been taken but a work around i'm using is:
from multiprocessing import Pool
t = None
def run(n):
return t.f(n)
class Test(object):
def __init__(self, number):
self.number = number
def f(self, x):
print x * self.number
def pool(self):
pool = Pool(2)
pool.map(run, range(10))
if __name__ == '__main__':
t = Test(9)
t.pool()
pool = Pool(2)
pool.map(run, range(10))
Output should be:
0
9
18
27
36
45
54
63
72
81
0
9
18
27
36
45
54
63
72
81
class Calculate(object):
# Your instance method to be executed
def f(self, x, y):
return x*y
if __name__ == '__main__':
inp_list = [1,2,3]
y = 2
cal_obj = Calculate()
pool = Pool(2)
results = pool.map(lambda x: cal_obj.f(x, y), inp_list)
There is a possibility that you would want to apply this function for each different instance of the class. Then here is the solution for that also
class Calculate(object):
# Your instance method to be executed
def __init__(self, x):
self.x = x
def f(self, y):
return self.x*y
if __name__ == '__main__':
inp_list = [Calculate(i) for i in range(3)]
y = 2
pool = Pool(2)
results = pool.map(lambda x: x.f(y), inp_list)
From http://www.rueckstiess.net/research/snippets/show/ca1d7d90 and http://qingkaikong.blogspot.com/2016/12/python-parallel-method-in-class.html
We can make an external function and seed it with the class self object:
from joblib import Parallel, delayed
def unwrap_self(arg, **kwarg):
return square_class.square_int(*arg, **kwarg)
class square_class:
def square_int(self, i):
return i * i
def run(self, num):
results = []
results = Parallel(n_jobs= -1, backend="threading")\
(delayed(unwrap_self)(i) for i in zip([self]*len(num), num))
print(results)
OR without joblib:
from multiprocessing import Pool
import time
def unwrap_self_f(arg, **kwarg):
return C.f(*arg, **kwarg)
class C:
def f(self, name):
print 'hello %s,'%name
time.sleep(5)
print 'nice to meet you.'
def run(self):
pool = Pool(processes=2)
names = ('frank', 'justin', 'osi', 'thomas')
pool.map(unwrap_self_f, zip([self]*len(names), names))
if __name__ == '__main__':
c = C()
c.run()
To implement multiprocessing in aws lambda we have two ways.
Note : Threadpool doesn't work in aws lambda
use the example solution which is provided by aws team
please use this link https://aws.amazon.com/blogs/compute/parallel-processing-in-python-with-aws-lambda/
use this package https://pypi.org/project/lambda-multiprocessing/
i have implemented my lambda function with both the solution and both is working fine can't share my code here but this 2 links will help you for sure.
i find 2 nd way more easy to implement.
There are also some libraries to make this easier, for example autothread (only for Python 3.6 and up):
import autothread
class calculate(object):
def run(self):
#autothread.multiprocessed()
def f(x: int):
return x*x
return f([1,2,3])
cl = calculate()
print(cl.run())
You can also take a look at lox.