与多处理池并行创建对象?

时间:2013-04-02 09:09:01

标签: python parallel-processing multiprocessing pickle

Python 2.7.3

我有一个包含数千个数据文件的文件夹。每个数据文件都被送到构造函数并进行大量处理。现在我正在遍历文件并按顺序处理它们:

class Foo:
    def __init__(self,file):
        self.bar = do_lots_of_stuff_with_numpy_and_scipy(file)

def do_lots_of_stuff_with_numpy_and_scipy(file):
    pass

def get_foos(dir):
    return [Foo(os.path.join(dir,file)) for file in os.listdir(dir)]

这很好用,但速度很慢。我想同时这样做。我试过了:

def parallel_get_foos(dir):
    p = Pool()
    foos = p.map(Foo, [os.path.join(dir,file) for file in os.listdir(dir)])
    p.close()
    p.join()
    return foos

if __name__ == "__main__":
    foos = parallel_get_foos(sys.argv[1])

但它只是出错了很多:

Process PoolWorker-7:
Traceback (most recent call last):
  File "/l/python2.7/lib/python2.7/multiprocessing/process.py", line 258, in _bootstrap
    self.run()
  File "/l/python2.7/lib/python2.7/multiprocessing/process.py", line 114, in run
    self._target(*self._args, **self._kwargs)
  File "/l/python2.7/lib/python2.7/multiprocessing/pool.py", line 99, in worker
    put((job, i, result))
  File "/l/python2.7/lib/python2.7/multiprocessing/queues.py", line 390, in put
    return send(obj)
PicklingError: Can't pickle <type 'function'>: attribute lookup __builtin__.function failed

我尝试过一个函数来返回对象,例如:

def get_foo(file):
    return Foo(file)

def parallel_get_foos(dir):
    ...
    foos = p.map(get_foo, [os.path.join(dir,file) for file in os.listdir(dir)])
    ...

但正如预期的那样我得到了同样的错误。

我已经阅读了大量类似的线程,试图解决类似这样的问题,但没有一个解决方案帮助了我。所以我感谢任何帮助!

编辑:

Bakuriu正确地推测我在do_lots_of_stuff方法中定义了一个非顶级函数。特别是,我的做法如下:

def fit_curve(data,degree):
    """Fits a least-square polynomial function to the given data."""
    sorted = data[data[:,0].argsort()].T
    coefficients = numpy.polyfit(sorted[0],sorted[1],degree)
    def eval(val,deg=degree):
        res = 0
        for coefficient in coefficients:
            res += coefficient*val**deg
            deg -= 1
        return res
    return eval

有没有让这个功能腌制?

1 个答案:

答案 0 :(得分:1)

你在做什么(至少,你在例子中显示的内容),实际上工作正常:

$mkdir TestPool
$cd TestPool/
$for i in {1..100}
> do
>     touch "test$i"
> done
$ls
test1    test18  test27  test36  test45  test54  test63  test72  test81  test90
test10   test19  test28  test37  test46  test55  test64  test73  test82  test91
test100  test2   test29  test38  test47  test56  test65  test74  test83  test92
test11   test20  test3   test39  test48  test57  test66  test75  test84  test93
test12   test21  test30  test4   test49  test58  test67  test76  test85  test94
test13   test22  test31  test40  test5   test59  test68  test77  test86  test95
test14   test23  test32  test41  test50  test6   test69  test78  test87  test96
test15   test24  test33  test42  test51  test60  test7   test79  test88  test97
test16   test25  test34  test43  test52  test61  test70  test8   test89  test98
test17   test26  test35  test44  test53  test62  test71  test80  test9   test99
$vi test_pool_dir.py
$cat test_pool_dir.py 
import os
import multiprocessing

class Foo(object):
    def __init__(self, fname):
        self.fname = fname   #or your calculations


def parallel_get_foos(directory):
    p = multiprocessing.Pool()
    foos = p.map(Foo, [os.path.join(directory, fname) for fname in os.listdir(directory)])
    p.close()
    p.join()
    return foos

if __name__ == '__main__':
    foos = parallel_get_foos('.')
    print len(foos)   #expected 101: 100 files plus this script

$python test_pool_dir.py 
101

版本信息:

$python --version
Python 2.7.3
$uname -a
Linux giacomo-Acer 3.2.0-39-generic #62-Ubuntu SMP Thu Feb 28 00:28:53 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux

我的猜测是你正在你所展示的代码示例中显示的内容。例如,在执行此操作时,我会收到类似于您的错误:

>>> import pickle
>>> def test():
...     def test2(): pass
...     return test2
... 
>>> import multiprocessing
>>> p = multiprocessing.Pool()
>>> p.map(test(), [1,2,3])
Exception in thread Thread-2:
Traceback (most recent call last):
  File "/usr/lib/python2.7/threading.py", line 551, in __bootstrap_inner
    self.run()
  File "/usr/lib/python2.7/threading.py", line 504, in run
    self.__target(*self.__args, **self.__kwargs)
  File "/usr/lib/python2.7/multiprocessing/pool.py", line 319, in _handle_tasks
    put(task)
PicklingError: Can't pickle <type 'function'>: attribute lookup __builtin__.function failed

这是显而易见的:

>>> pickle.dumps(test())
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python2.7/pickle.py", line 1374, in dumps
    Pickler(file, protocol).dump(obj)
  File "/usr/lib/python2.7/pickle.py", line 224, in dump
    self.save(obj)
  File "/usr/lib/python2.7/pickle.py", line 286, in save
    f(self, obj) # Call unbound method with explicit self
  File "/usr/lib/python2.7/pickle.py", line 748, in save_global
    (obj, module, name))
pickle.PicklingError: Can't pickle <function test2 at 0x7fad15fc2938>: it's not found as __main__.test2

pickle的文件说明:

  

可以腌制以下类型:

     
      
  • NoneTrueFalse
  •   
  • 整数,长整数,浮点数,复数
  •   
  • 普通和Unicode字符串
  •   
  • tuple s,list s,set s和仅包含可选对象的词典
  •   
  • 函数在模块顶层定义
  •   
  • 在模块顶层定义的内置函数
  •   
  • 在模块顶层定义的类
  •   
  • 此类的实例__dict__或调用__getstate__()的结果是可选的(请参阅pickle协议一节)   详情)。
  •   

并继续:

  

请注意,功能(内置和用户定义)被“完全”腌制   合格的“名称参考,而不是价值。这意味着只有   功能名称被腌制,以及模块的名称   函数定义为。函数的代码,也不是函数的代码   功能属性是pickle。因此,定义模块必须是   可以在unpickling环境中导入,并且模块必须包含   命名对象,否则将引发异常。