PropertyMock 在每次调用后更改值

PropertyMock with changing values after every call

我正在尝试模拟 psycopg2 并且需要处理 cursor.description

  # run query 1
  column_names = [desc[0] for desc in cursor.description]
  # run query 2
  # NB cursor.description now returns different value
  column_names = [desc[0] for desc in cursor.description]

我可以通过 PropertyMock 做到这一点:

  type(fake_cursor).description = PropertyMock(return_value=descriptions[0])

但这并没有解决我的问题,因为被测系统做了两次查询,第二次的描述会有所不同。

如何在第二次调用时更改模拟的 return 值?

我试过了:

cursor_execute_call_count = 0

def handle_cursor_execute(arg1, arg2):
    cursor_execute_call_count = cursor_execute_call_count + 1
    type(fake_cursor).description = PropertyMock(return_value=descriptions[cursor_execute_call_count])

fake_cursor.execute = Mock(side_effect=handle_cursor_execute)

但我明白了

E     UnboundLocalError: local variable 'cursor_execute_call_count' referenced before assignment

(这对我来说没有意义。肯定存在奇怪的范围界定问题。)

我最终确实使用了这段代码,但似乎必须有更好的方法:

def handle_cursor_execute(arg1, arg2):
  description = descriptions.pop()
  type(fake_cursor).description = PropertyMock(return_value=description)

根据说明进行编辑

我想你可以实现一个迭代器:

class MyDescription():

    descriptions = iter([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

    def __iter__(self):
        for description_list in self.descriptions:
            for description in description_list:
                yield description
            raise StopIteration
type(fake_cursor).description = PropertyMock(return_value=MyDescription())

所以它看起来像这样:

>>> z = MyDescription()
>>> [i for i in z]
[1, 2, 3]
>>> [i for i in z]
[4, 5, 6]
>>> [i for i in z]
[7, 8, 9]
>>>