如何修补异步类方法？

Question

我正在解决以下问题，我有一个类，一个我想模拟补丁的异步方法：

class ExampleClass:
    async def asy_method(self, param):
        return await some_coroutine(self, param)

example_instance = ExampleClass()

我想专门修补像这样的调用

await example_instance.asy_method('test_param')

通常我会用

mocker.patch('ExampleClass.asy_method', new_callable=AsyncMock)

其中 mocker 是 pytest-mock 插件夹具，而 AsyncMock 具有以下形式

class AsyncMock(mock.MagicMock):
    async def __call__(self, *args, **kwargs):
         return super(AsyncMock, self).__call__(*args, **kwargs)

这会给我一个 Mock 对象，它的行为就像一个调用时的协程。问题是，我想访问self传递给该方法的属性。self仅在您设置时才传递给模拟对象 autospec=True（另请参阅有关修补未绑定方法的 Python 文档），您不能将其与new_callable.

有谁知道如何解决这个问题？

Answer 1

实际上，您不能将自动指定和新的可调用对象混合使用。相反，自动指定方法，然后替换side_effect属性，给它一个AsyncMock()实例：

from unittest import mock


def configure_coroutine_mock(mock_function, klass=AsyncMock):
    """Make an autospecced async function return a coroutine mock"""
    mock_function.side_effect = AsyncMock()
    # mark the side effect as a child of the original mock object
    # so transitive access is recorded on the parent mock too. This is 
    # what .return_value does normally
    mock._check_and_set_parent(
        mock_function.mock, mock_function.side_effect,
        None, '()')
    return mock_asy_method.side_effect


with mocker.patch('ExampleClass.asy_method', autospec=True) as mock_asy_method:
    configure_coroutine_mock(mock_asy_method)

因为AsyncMock()是一个可调用对象，所以每次被调用时mock_asy_method都会调用它，并将参数传递给对象。然后使用该调用的结果从 返回mock_asy_method()：

>>> from unittest import mock
>>> class ExampleClass:
...     async def asy_method(self, param):
...         return await some_coroutine(self, param)
...
>>> example_instance = ExampleClass()
>>> with mock.patch('__main__.ExampleClass.asy_method', autospec=True) as mock_asy_method:
...     configure_coroutine_mock(mock_asy_method)
...     print(example_instance.asy_method('foo'))  # call to patched class coroutine
...     print(mock_asy_method.mock_calls)          # calls are recorded
...
<AsyncMock name='asy_method()' id='4563887496'>
<coroutine object AsyncMock.__call__ at 0x1100780f8>
[call(<__main__.ExampleClass object at 0x10ffac1d0>, 'foo')]

如您所见，self参数和参数都记录在调用中，因为它mock_asy_method是一个正确指定的函数。

当然，只有在AsyncMock()实际等待返回的调用结果时，我们才会看到该调用也被记录了：

>>> with mock.patch('__main__.ExampleClass.asy_method', autospec=True) as mock_asy_method:
...     configure_coroutine_mock(mock_asy_method)
...     loop = asyncio.get_event_loop()
...     coro = example_instance.asy_method('foo')
...     loop.run_until_complete(coro)
...     print(mock_asy_method.mock_calls)
...     
<AsyncMock name='asy_method()' id='4564408920'>
<AsyncMock name='asy_method()()' id='4564999360'>    
[call(<__main__.ExampleClass object at 0x10ffac1d0>, 'foo'),
 call()(<__main__.ExampleClass object at 0x10ffac1d0>, 'foo')]