通常来说,实现上下文管理器,需要编写一个带有__enter__和 __exit__的类,类似这样:
class ListTransaction: def __init__(self, orig_list): self.orig_list = orig_list self.working = list(orig_list) def __enter__(self): return self.working def __exit__(self, exc_type, exc_val, exc_tb): self.orig_list[:] = self.working
然而,在contextlib模块中,还提供了@contextmanager装饰器,将一个生成器函数当成上下文管理器使用,上面的代码在大部分,是与下面的代码等效的。
本文的list_transaction函数的代码来自:《Python Cookbook》 9.22 以简单的方式定义上下文管理器
from contextlib import contextmanager
@contextmanager def list_transaction(orig_list): working = list(orig_list) yield working orig_list[:] = working
先逐一分析上面的代码:
- 因为list是可变类型,所以通过list(orig_list),对值进行复制,创建一个新的list,即working。
- 以yield为分隔,在yield之前的代码,包括yield working,会在contextmanager装饰器的__enter__方法中被调用
- 代码在执行到yield时暂停,同时yield working,会将working产出。yield产出的值,作为__enter__的返回值,赋值给as之后的变量
- 当with块的代码执行完成后, 上下文管理器会在yield处恢复,继续执行yield之后的代码。
- yield 之后的代码,则在contextmanager装饰器中的__exit__方法中被调用
测试代码如下:
当执行过程中,没有引发异常时,执行正常,输出 [1, 2, 3, 4, 5]
items_1 = [1, 2, 3] with list_transaction(items_1) as working_1: working_1.append(4) working_1.append(5) print(items_1)
当执行过程中,引发异常时,yield后的代码不会执行,orig_list不会被修改。从而实现事务的效果,orig_list仍是 [1, 2, 3]
items_2 = [1, 2, 3] try: with list_transaction(items_2) as working_2: working_2.append(4) working_2.append(5) raise RuntimeError('oops') except Exception as ex: print(ex) finally: print(items_2)
上下文管理器类与@contextmanager中最大的区别在于对异常的处理。
分析contextmanager的源码可知,@contextmanager装饰器的本质是实例化一个_GeneratorContextManager对象。
def contextmanager(func): @wraps(func) def helper(*args, **kwds): return _GeneratorContextManager(func, args, kwds) return helper
进一步查看_GeneratorContextManager源码,可知_GeneratorContextManager实现的是一个上下文管理器对象
class _GeneratorContextManager(ContextDecorator): """Helper for @contextmanager decorator.""" def __init__(self, func, args, kwds): self.gen = func(*args, **kwds) self.func, self.args, self.kwds = func, args, kwds # Issue 19330: ensure context manager instances have good docstrings doc = getattr(func, "__doc__", None) if doc is None: doc = type(self).__doc__ self.__doc__ = doc # Unfortunately, this still doesn't provide good help output when # inspecting the created context manager instances, since pydoc # currently bypasses the instance docstring and shows the docstring # for the class instead. # See http://bugs.python.org/issue19404 for more details. def _recreate_cm(self): # _GCM instances are one-shot context managers, so the # CM must be recreated each time a decorated function is # called return self.__class__(self.func, self.args, self.kwds) def __enter__(self): try: return next(self.gen) except StopIteration: raise RuntimeError("generator didn't yield") from None def __exit__(self, type, value, traceback): if type is None: try: next(self.gen) except StopIteration: return else: raise RuntimeError("generator didn't stop") else: if value is None: # Need to force instantiation so we can reliably # tell if we get the same exception back value = type() try: self.gen.throw(type, value, traceback) raise RuntimeError("generator didn't stop after throw()") except StopIteration as exc: # Suppress StopIteration *unless* it's the same exception that # was passed to throw(). This prevents a StopIteration # raised inside the "with" statement from being suppressed. return exc is not value except RuntimeError as exc: # Likewise, avoid suppressing if a StopIteration exception # was passed to throw() and later wrapped into a RuntimeError # (see PEP 479). if exc.__cause__ is value: return False raise except: # only re-raise if it's *not* the exception that was # passed to throw(), because __exit__() must not raise # an exception unless __exit__() itself failed. But throw() # has to raise the exception to signal propagation, so this # fixes the impedance mismatch between the throw() protocol # and the __exit__() protocol. # if sys.exc_info()[1] is not value: raise
简要分析实现的代码:
__enter__方法:
- self.gen = func(*args, **kwds) 获取生成器函数返回的生成器,并赋值给self.gen
- with代码块进入__enter__方法时,调用生成器的__next__方法,使代码执行到yield处暂停
- 将yield产出的值作为__enter__的返回值
- 因为__enter__方法只会执行一次,如果第一次调用生成器的__next__方法,就抛出StopIteration异常,说明生成器存在问题,则抛出RuntimeError
__exit__方法:
正常执行的情况:
- def __exit__(self, type, value, traceback)接收三个参数,第一个参数是异常类,第二个参数是异常对象,第三个参数是trackback对象
- 如果with内的代码执行正常,没有抛出异常,则上述三个参数都为None
- __exit__代码中首先对type是否None进行判断,如果type为None,说明with代码内部执行正常,所以调用生成器的__next__方法。此时生成器在yield处恢复运行,继续执行yield之后的代码
- 正常情况下,调用__next__方法,迭代应结束,抛出StopIteration异常;如果没有抛出StopIteration异常,说明生成器存在问题,则抛出RuntimeError
出现异常的情况:
- 如果type类型不为None,说明在with代码内部执行时出现异常。如果异常对象value为None,则强制使用异常类实例化一个新的异常对象,并赋值给value
- 使用throw方法,将异常对象value传递给生成器函数,此时生成器在yield处恢复执行,并接收到异常信息
- 通常情况下,yield语句应该在try except代码块中执行,用于捕获__exit__方法传递给生成器的异常信息,并进行处理
- 如果生成器函数可以处理异常,迭代完成后,自动抛出StopIteration。
- __exit__ 捕获并压制StopIteration,除非with内的代码也抛出了StopIteration。return exc is not value,exc是捕获到的StopIteration异常实例,value是with内代码执行时抛出的异常。在__exit__方法中,return True告诉解释器异常已经处理,除此以外,所有的异常都会向上冒泡。
- 如果生成器没有抛出StopIteration异常,说明迭代没有正常结束,则__exit__方法抛出RuntimeError,同样的,除非with代码块内部也抛出RuntimeError,否则RuntimeError会在__exit__中被捕获并且压制。
所以,以类的方式实现的上下文管理器,在引发异常时,__exit__方法内的代码仍会正常执行;
而以生成器函数实现的上下文管理器,在引发异常时,__exit__方法会将异常传递给生成器,如果生成器无法正确处理异常,则yield之后的代码不会执行。
所以,大部分情况下,yield都必须在try...except中,除非设计之初就是让yield之后的代码在with代码块内部出现异常时不执行。
测试代码:
以类的方式实现上下文管理器,当没有引发异常时, # 其执行结果与@contextmanager装饰器装饰器的上下文管理器函数相同,输出 [1, 2, 3, 4, 5]
items_3 = [1, 2, 3] with ListTransaction(items_3) as working_3: working_3.append(4) working_3.append(5) print(items_3)
当执行代码过程中引发异常时,即使没有对异常进行任何处理,__exit__方法也会正常执行,对self.orig_list进行修改(python是引用传值,而list是可变类型,对orig_list的任何引用的修改,都会改变orig_list的值),所以输出结果与没有引发异常时相同:[1, 2, 3, 4, 5]
items_4 = [1, 2, 3] try: with ListTransaction(items_4) as working_4: working_4.append(4) working_4.append(5) raise RuntimeError('oops') except Exception as ex: print(ex) finally: print(items_4)
完整代码:https://github.com/blackmatrix7/python-learning/blob/master/class_/contextlib_.py