python 为函数加上自动retry 的装饰器

"""
    :param delays:      An iterable object the defines num of seconds between reties, also indicating num of retries. 
                        Default: retry 3 times, waiting 1/5/30 seconds.
    :param exceptions:  Specifies exceptions that trigger retry. Otherwise no actions will be taken. Default: all exceptions.
    :param report:      An callable object, e.g print or logger method. Default: report nothing.
"""
def retry(delays=(3, 10, 15), exceptions=(Exception, ), report=lambda *args: None):
    def wrapper(function):
        @functools.wraps(function)
        def wrapped(*args, **kwargs):
            problems = []
            for delay in itertools.chain(delays, [None]):
                try:
                    return function(*args, **kwargs)
                except exceptions as problem:
                    problems.append(problem)
                    if delay is None:
                        report("Function {func} failed after {num} retries. Exceptions: {p}".format(func=function.__name__, num=len(delays), p=problems))
                        raise
                    else:
                        report("Function {func} fails of {expt_name}({expt_msg}). Will retry in {delay} second(s)."\
                               .format(func=function.__name__, expt_name=type(problem).__name__, expt_msg=problem, delay=delay))
                        time.sleep(delay)
        return wrapped
    return wrapper

上面这个装饰器函数来自某位大神之手，先膜拜一下，它能自动为被装饰的函数加上retry功能。

根据上面的代码，会retry三次，每次间隔3，10，15秒。

@functools.wraps(function)的作用是使被装饰的函数保留原有的函数名字和函数doc。

 

这个函数设计的很巧妙，比如 report=lambda *args: None

作者定义了"返回为None的函数"作为“report”参数的默认值。