Pytest结合pytest-mock实现依赖对象方法调用监听与传参校验

class Work:
    pass

class Engineer:
   def delegate(work: Work):
       print('Doing all the hard work')

class Manager:
   def __init__(self, engineer: Engineer):
       self.engineer = engineer

   def perform(work: Work):
       self.engineer.delegate(work)

测试Manager类的perform()方法，验证engineer.delegate()方法是否被传入了正确的参数，预期测试逻辑伪代码如下：

# pseudo 
work = mock(Work)
engr = mock(Engineer)

mgr = spy(Manager)
mgr.perform(work)

verify(engr.delegate).times(1).capture(args)
assert args == work

首先注意：原代码中Engineer.delegate、Manager.perform作为实例方法漏写了首个self参数，直接运行会报错，需要先修正方法签名：

# 修正后的实例方法
class Engineer:
   def delegate(self, work: Work):
       print('Doing all the hard work')

class Manager:
   def __init__(self, engineer: Engineer):
       self.engineer = engineer

   def perform(self, work: Work):
       self.engineer.delegate(work)

测试代码直接使用pytest-mock自带的mocker fixture即可实现，不需要额外spy Manager对象——单元测试中被测类使用真实实例，仅替换其依赖为mock对象即可：

def test_manager_perform_pass_correct_param(mocker):
    # 创建mock依赖
    mock_work = mocker.Mock(spec=Work)
    mock_engineer = mocker.Mock(spec=Engineer)

    # 初始化真实被测对象，注入mock依赖
    manager = Manager(engineer=mock_engineer)

    # 执行被测逻辑
    manager.perform(mock_work)

    # 校验：delegate仅被调用1次，且入参为传入的work对象
    mock_engineer.delegate.assert_called_once_with(mock_work)

逻辑对应说明 #

• mocker.Mock(spec=类) 对应伪代码中的mock(类)，spec参数会限制mock对象仅存在原类的属性和方法，避免拼写错误导致测试无效
• 不需要spy Manager：我们要验证的是Manager的真实执行逻辑，spy反而会替换真实逻辑，不符合单元测试要求
• assert_called_once_with(目标参数) 同时覆盖了伪代码中的两个校验点：校验调用次数为1、校验入参与预期完全一致

内容的提问来源于stack exchange，提问作者Keerthi