上篇博文介绍了常见需要进行请求重试的场景，本篇博文试着剖析有名的python第三方库retrying源码。
在剖析其源码之前，有必要讲一下retrying的用法，方便理解。
安装：
pip install retrying
或者
easy_install retrying
一些用法实例如下:
#example 1 from retrying import retry @retry def never_give_up_never_surrender(): print "一直重试且两次重试之间无需等待"
#example 2 from retrying import retry @retry(stop_max_attempt_number=7) def stop_after_7_attempts(): print "重试七次后停止"
#example 3 from retrying import retry @retry(stop_max_delay=10000) def stop_after_10_s(): print "十秒之后停止重试"
#example 4 from retrying import retry @retry(wait_fixed=2000) def wait_2_s(): print "每次重试间隔两秒"
#example 5 from retrying import retry @retry(wait_random_min=1000, wait_random_max=2000) def wait_random_1_to_2_s(): print "每次重试随机等待1到2秒"
#example 6 from retrying import retry @retry(wait_exponential_multiplier=1000, wait_exponential_max=10000) def wait_exponential_1000(): print "指数退避，每次重试等待 2^x * 1000 毫秒，上限是10秒，达到上限后每次都等待10秒"
#example 7 def retry_if_io_error(exception): """Return True if we should retry (in this case when it's an IOError), False otherwise""" return isinstance(exception, IOError) @retry(retry_on_exception=retry_if_io_error) def might_io_error(): print "IO异常则重试，并且将其它异常抛出" @retry(retry_on_exception=retry_if_io_error, wrap_exception=True) def only_raise_retry_error_when_not_io_error(): print "IO异常则重试，并且将其它异常用RetryError对象包裹"
#exampe 8，根据返回结果判断是否重试 def retry_if_result_none(result): """Return True if we should retry (in this case when result is None), False otherwise""" return result is None @retry(retry_on_result=retry_if_result_none) def might_return_none(): print "若返回结果为None则重试"
上面八个例子是retrying的用法,只需在要重试的方法上加上@retry注解，并以相应的条件为参数即可,那么@retry背后到底是如何实现的呢?下面给出@retry注解实现的方法。
1 #装饰器模式,对需要重试的函数,利用retry注解返回 2 def retry(*dargs, **dkw): 3 """ 4 Decorator function that instantiates the Retrying object 5 @param *dargs: positional arguments passed to Retrying object 6 @param **dkw: keyword arguments passed to the Retrying object 7 """ 8 # support both @retry and @retry() as valid syntax 9 #当用法为@retry不带括号时走这条路径,dargs[0]为retry注解的函数,返回函数对象wrapped_f 10 if len(dargs) == 1 and callable(dargs[0]): 11 def wrap_simple(f): 12 13 @six.wraps(f)#注解用于将函数f的签名复制到新函数wrapped_f 14 def wrapped_f(*args, **kw): 15 return Retrying().call(f, *args, **kw) 16 17 return wrapped_f 18 19 return wrap_simple(dargs[0]) 20 21 else:#当用法为@retry()带括号时走这条路径，返回函数对象wrapped_f 22 def wrap(f): 23 24 @six.wraps(f)#注解用于将函数f的签名复制到新函数wrapped_f 25 def wrapped_f(*args, **kw): 26 return Retrying(*dargs, **dkw).call(f, *args, **kw) 27 28 return wrapped_f 29 30 return wrap
当用@retry标记函数时，例如实例1，其实执行了
never_give_up_never_surrender = retry(never_give_up_never_surrender)
此时的never_give_up_never_surrender函数实际上是10-19行返回的wrapped_f函数，后续对never_give_up_never_surrender函数的调用都是调用的14行的wrapped_f函数。
当使用@retry()或者带参数的@retry(params)时，如实例2,实际执行了:
stop_after_7_attempts = retry(stop_max_attempt_number)(stop_after_7_attempts)
此时的stop_after_7_attempts函数实际上是22-29行的wrapped_f函数，后续对stop_after_7_attempts函数的调用都是对25行的wrapped_f函数调用。
可以看到实际上@retry将对需要重试的函数调用转化为对Retrying类中call函数的调用，重试逻辑也在这个函数实现,实现对逻辑代码的无侵入，代码如下:
1 def call(self, fn, *args, **kwargs): 2 start_time = int(round(time.time() * 1000)) 3 attempt_number = 1 4 while True: 5 #_before_attempts为@retry传进来的before_attempts，在每次调用函数前执行一些操作 6 if self._before_attempts: 7 self._before_attempts(attempt_number) 8 9 try:#Attempt将函数执行结果或者异常信息以及执行次数作为内部状态,用True或False标记是内部存的值正常执行结果还是异常 10 attempt = Attempt(fn(*args, **kwargs), attempt_number, False) 11 except: 12 tb = sys.exc_info()#获取异常堆栈信息，sys.exc_info()返回type(异常类型), value(异常说明), traceback(traceback对象，包含更丰富的信息) 13 attempt = Attempt(tb, attempt_number, True) 14 15 if not self.should_reject(attempt):#根据本次执行结果或异常类型判断是否应该停止 16 return attempt.get(self._wrap_exception) 17 18 if self._after_attempts:#_after_attempts为@retry传进来的after_attempts，在每次调用函数后执行一些操作 19 self._after_attempts(attempt_number) 20 21 delay_since_first_attempt_ms = int(round(time.time() * 1000)) - start_time 22 if self.stop(attempt_number, delay_since_first_attempt_ms):#根据重试次数和延迟判断是否应该停止 23 if not self._wrap_exception and attempt.has_exception: 24 # get() on an attempt with an exception should cause it to be raised, but raise just in case 25 raise attempt.get() 26 else: 27 raise RetryError(attempt) 28 else:#不停止则等待一定时间，延迟时间根据wait函数返回值和_wait_jitter_max计算 29 sleep = self.wait(attempt_number, delay_since_first_attempt_ms) 30 if self._wait_jitter_max: 31 jitter = random.random() * self._wait_jitter_max 32 sleep = sleep + max(0, jitter) 33 time.sleep(sleep / 1000.0) 34 35 attempt_number += 1 #进行下一轮重试
9-13行将函数执行返回结果或异常存入Attempt对象attempt中,Attempt类如下:
class Attempt(object): """ An Attempt encapsulates a call to a target function that may end as a normal return value from the function or an Exception depending on what occurred during the execution. """ #value值为函数返回结果或异常，根据has_exception判断 def __init__(self, value, attempt_number, has_exception): self.value = value self.attempt_number = attempt_number self.has_exception = has_exception #返回函数执行结果或异常，并根据wrap_exception参数对异常用RetryError包裹 def get(self, wrap_exception=False): """ Return the return value of this Attempt instance or raise an Exception. If wrap_exception is true, this Attempt is wrapped inside of a RetryError before being raised. """ if self.has_exception: if wrap_exception: raise RetryError(self) else:#重新构造原异常抛出 six.reraise(self.value[0], self.value[1], self.value[2]) else: return self.value def __repr__(self): if self.has_exception: return "Attempts: {0}, Error:\n{1}".format(self.attempt_number, "".join(traceback.format_tb(self.value[2]))) else: return "Attempts: {0}, Value: {1}".format(self.attempt_number, self.value)
15行根据should_reject函数的返回值判断是否停止重试，代码如下:
def should_reject(self, attempt): reject = False #假如异常在retry_on_exception参数中返回True，则重试,默认不传异常参数时，发生异常一直重试 if attempt.has_exception: reject |= self._retry_on_exception(attempt.value[1]) else:#假如函数返回结果在retry_on_result参数函数中为True，则重试 reject |= self._retry_on_result(attempt.value) return reject
22行根据重试次数和延迟判断是否应该停止重试，self.stop的赋值代码在构造函数中，代码片段如下:
stop_funcs = [] if stop_max_attempt_number is not None: stop_funcs.append(self.stop_after_attempt) if stop_max_delay is not None: stop_funcs.append(self.stop_after_delay) if stop_func is not None: self.stop = stop_func elif stop is None:#执行次数和延迟任何一个达到限制则停止 self.stop = lambda attempts, delay: any(f(attempts, delay) for f in stop_funcs) else: self.stop = getattr(self, stop)
def stop_after_attempt(self, previous_attempt_number, delay_since_first_attempt_ms): """Stop after the previous attempt >= stop_max_attempt_number.""" return previous_attempt_number >= self._stop_max_attempt_number def stop_after_delay(self, previous_attempt_number, delay_since_first_attempt_ms): """Stop after the time from the first attempt >= stop_max_delay.""" return delay_since_first_attempt_ms >= self._stop_max_delay
29-33行等待一段时间再次重试，其中延迟时间重点是根据29行的wait函数计算，wait函数在构造函数中赋值，代码片段如下:
wait_funcs = [lambda *args, **kwargs: 0] if wait_fixed is not None: wait_funcs.append(self.fixed_sleep) if wait_random_min is not None or wait_random_max is not None: wait_funcs.append(self.random_sleep) if wait_incrementing_start is not None or wait_incrementing_increment is not None: wait_funcs.append(self.incrementing_sleep) if wait_exponential_multiplier is not None or wait_exponential_max is not None: wait_funcs.append(self.exponential_sleep) if wait_func is not None: self.wait = wait_func elif wait is None:#返回几个函数的最大值,作为等待时间 self.wait = lambda attempts, delay: max(f(attempts, delay) for f in wait_funcs) else: self.wait = getattr(self, wait)
其中最值得研究的是指数退避延迟时间计算方法，函数为exponential_sleep,代码如下:
def exponential_sleep(self, previous_attempt_number, delay_since_first_attempt_ms): exp = 2 ** previous_attempt_number result = self._wait_exponential_multiplier * exp #延迟时间为_wait_exponential_multiplier*2^x if result > self._wait_exponential_max:#假如大于退避上限_wait_exponential_max,则result为上限值 result = self._wait_exponential_max if result < 0: result = 0 return result