How to obtain the results from a pool of threads in python?

Question:

I have searched here about how to do threading in python, but by far i haven’t been able to get the answer i need.
I’m not very familiar with the Queue and Threading python classes and for that reason some of the answers present here makes no sense at all to me.

I want to create a pool of threads which i can give different task and when all of them have ended get the result values and process them.
So far i have tried to do this but i’m not able to get the results. The code i have written is:

from threading import Thread
from Queue import Queue

class Worker(Thread):
    """Thread executing tasks from a given tasks queue"""
    def __init__(self, tasks):
        Thread.__init__(self)
        self.tasks = tasks
        self.daemon = True
        self.result = None
        self.start()
    def run(self):
        while True:
            func, args, kargs = self.tasks.get()
            try:
                self.result = func(*args, **kargs)
            except Exception, e:
                print e
            self.tasks.task_done()
    def get_result(self):
        return self.result

class ThreadPool:
    """Pool of threads consuming tasks from a queue"""
    def __init__(self, num_threads):
        self.tasks = Queue(num_threads)
        self.results = []
        for _ in range(num_threads):
            w = Worker(self.tasks)
            self.results.append(w.get_result())
    def add_task(self, func, *args, **kargs):
        """Add a task to the queue"""
        self.tasks.put((func, args, kargs))
    def wait_completion(self):
        """Wait for completion of all the tasks in the queue"""
        self.tasks.join()
    def get_results(self):
        return self.results

def foo(word, number):
    print word*number
    return number

words = ['hello', 'world', 'test', 'word', 'another test']
numbers = [1,2,3,4,5]
pool = ThreadPool(5)
for i in range(0, len(words)):
    pool.add_task(foo, words[i], numbers[i])

pool.wait_completion()
results = pool.get_results()
print results

The output prints the strings with word given times the number given but the results list is full with None values, so where i should put the return values of the func.

Or the easy way is to create a list where i fill the Queue and add a dictionary or some variable to store the result as an argument to my function, and after the task is added to the Queue add this result argument to a list of results:

def foo(word, number, r):
    print word*number
    r[(word,number)] = number
    return number

words = ['hello', 'world', 'test', 'word', 'another test']
numbers = [1,2,3,4,5]
pool = ThreadPool(5)
results = []
for i in range(0, len(words)):
    r = {}
    pool.add_task(foo, words[i], numbers[i], r)
    results.append(r)
print results
Asked By: Rafael Rios

||

Answers:

In Python 3.x, you can use concurrent.futures.ThreadPoolExecutor to do this, rather than rolling your own.

Python 2.x actually has a built-in thread pool you can use as well, its just not well documented:

from multiprocessing.pool import ThreadPool

def foo(word, number):
    print (word * number)
    r[(word,number)] = number
    return number

words = ['hello', 'world', 'test', 'word', 'another test']
numbers = [1,2,3,4,5]
pool = ThreadPool(5)
results = []
for i in range(0, len(words)):
    results.append(pool.apply_async(foo, args=(words[i], numbers[i])))

pool.close()
pool.join()
results = [r.get() for r in results]
print results

Or (using map instead of apply_async):

from multiprocessing.pool import ThreadPool

def foo(word, number):
    print word*number
    return number

def starfoo(args):
    """ 

    We need this because map only supports calling functions with one arg. 
    We need to pass two args, so we use this little wrapper function to
    expand a zipped list of all our arguments.
    
    """    
    return foo(*args)

words = ['hello', 'world', 'test', 'word', 'another test']
numbers = [1,2,3,4,5]
pool = ThreadPool(5)
# We need to zip together the two lists because map only supports calling functions
# with one argument. In Python 3.3+, you can use starmap instead.
results = pool.map(starfoo, zip(words, numbers))
print results

pool.close()
pool.join()
Answered By: dano