Python : Parallel execution of function

Question

I would like to execute set of tasks in parallel. I have defined a function in a class which takes the parameter and executes the operation based on parameter. Class structure is like below.

from threading import Thread
from concurrent.futures import *
class Test(object):

  def process_dataframe(self,id:int):
    print(id*id)


  def run_task(self):
    thd = []
    for i in range(1,10): 
      thd.append( "self.process_dataframe({0})".format(i))
    self.run_functions_in_parallel(thd)

  def run_functions_in_parallel(self,fns)->bool:
    def wrap_function(self,fnToCall):
      try:
        eval(fnToCall)
        return ("0")
      except Exception as e:
        return "{0}".format(e)

    thd = []
    isError = False
    executor = ThreadPoolExecutor(max_workers=len(fns))
    errorMessage = ""

    for fn in fns:     
      t = executor.submit(wrap_function,self,fn)
      thd.append(t)

    for td in thd:
      ret = td.result()
      if ret != "0":
        isError = True
        errorMessage = errorMessage + """
        """ + ret
    if isError == True:
      print (errorMessage)
      raise Exception (errorMessage)
    else:
      return True


d=Test()
d.run_task()

I have managed to make it work and tasks are executing properly. I am wondering whether there is better/simpler way to accomplish the same. I would like to keep run_functions_in_parallel method generic so that it can be used as common method in a module.

If you really want parallelism, then you must be using multiprocessing — sagarr
– sagarr, Commented Dec 25, 2019 at 8:36
It is always a bad idea to use a magic string as error indicator. Try a function with content {}[0]. — Daniel
– Daniel, Commented Dec 25, 2019 at 8:53
Don't use eval. You can just use function handles, like submit also does. — Daniel
– Daniel, Commented Dec 25, 2019 at 8:55

Daniel · Accepted Answer · 2019-12-25 09:17:08Z

You don't need to use a wrapper, since ThreadPoolExecutor catches errors in a better way. A function, that always returns True or raises an error, don't need a return value, but if you have functions with return values, you want to call in parallel, you should return their results. It is a bad idea to use a magic string as indicator for errors. format(e) of a KeyError: 0 also leads to "0". Better use a unique value, like None in our case. Don't use eval if you don't have to. In your case, you can use partial. Don't use a to large value for max_workers.

from functools import partial
from concurrent.futures import ThreadPoolExecutor

class Test(object):
    def process_dataframe(self, id):
        print(id*id)

    def run_task(self):
        functions = []
        for i in range(1,10): 
            functions.append(partial(self.process_dataframe, i))
        self.run_functions_in_parallel(functions)

    def run_functions_in_parallel(self, functions, max_workers=8):
        executor = ThreadPoolExecutor(max_workers=max_workers)
        futures = [
            executor.submit(function)
            for function in functions
        ]

        errors = []
        results = []
        for future in futures:
            try:
                result = future.result()
            except Exception as e:
                errors.append(e)
            else:
                results.append(result)
        if errors:
            raise Exception(errors)
        return results

d = Test()
d.run_task()

Collectives™ on Stack Overflow

Python : Parallel execution of function

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related