python - 替代 __del__ 来获得最终结果并终止类中的进程？

Question

这个虚拟类与我当前的项目类似:

class EndProcess(object):

    def __init__(self):
        self._Q = multiprocessing.Queue()

        self.p1 = multiprocessing.Process(target=self.worker, args=("p1",))
        self.p2 = multiprocessing.Process(target=self.worker, args=("p2",))
        self.p3 = multiprocessing.Process(target=self.worker, args=("p3",))
        self.p1.start()
        self.p2.start()
        self.p3.start()


    def worker(self, name):
        while True:
            rnd = random.randint(0, 100)
            self._Q.put((name, rnd))
            print(name, rnd)
            time.sleep(1)


    def __del__(self):
        # "log" final state of Queue
        while not self._Q.empty():
            print(self._Q.get())

        # free resources
        ...

        # clean up workers
        self.p1.terminate()
        self.p2.terminate()
        self.p3.terminate()


if __name__ == "__main__":
    ep = EndProcess()

问题是 __del__ 不能保证总是被调用，特别是在重要的时间点，因为 __del__ 仅在引用计数时被 Python 的垃圾收集器调用对象的值达到零。

我如何确保资源始终被释放并且工作人员始终被终止？

Answer 1

How can I assure that resources always get set free and the workers always get terminated?

您可以通过实现上下文管理器方法 __enter__ 来实现此目的和 __exit__ 。然后，您可以将 EndProcess 类与上下文管理器语句 with 一起使用。这样，即使发生错误，您的清理逻辑也会执行:

class EndProcess(object):
    def __init__(self):
        self._Q = multiprocessing.Queue()
        self.p1 = multiprocessing.Process(target=self.worker, args=("p1",))
        self.p2 = multiprocessing.Process(target=self.worker, args=("p2",))
        self.p3 = multiprocessing.Process(target=self.worker, args=("p3",))
        self.p1.start()
        self.p2.start()
        self.p3.start()

    def worker(self, name):
        while True:
            rnd = random.randint(0, 100)
            self._Q.put((name, rnd))
            print(name, rnd)
            time.sleep(1)

    def __enter__(self):
        # Add your setup logic here. Initialize any data structures.
        # set any threads, etc.
        pass

    def __exit__(self, exc_type, exc_value, traceback):
        # The three arguments to `__exit__` describe the exception
        # caused the `with` statement execution to fail. If the `with`
        # statement finishes without an exception being raised, these
        # arguments will be `None`.
        #
        # These arguments may be useful for things such as logging, or 
        # debugging.
        while not self._Q.empty():
            print(self._Q.get())
        ...
        self.p1.terminate()
        self.p2.terminate()
        self.p3.terminate()
        ...

现在您可以在上下文管理器语句中使用 EndProcess 实例:

end_process = EndProcess()
...
with end_process as ep:
    # Use ep like you would normally. `__enter__` and `__exit__` will be
    # automatically.
    pass