[Python] Difference between mock.patch.dict as function decorator and context manager in...

I have a configuration module, config with data config.data where I want to add a value for testing. Ideally, I want to use mock.patch.dict as a context manager because the value is a class attribute and I need to use self.test_value. The problem is it doesn't work. The dict patching only works when used as a decorator, but not when used as a context manager and I can't figure out why.

This is more or less what I have:

This works:

@mock.patch.dict("config.data", {"test": 123})
@mock.patch("config.load_data", return_value=None, autospec=True)
def run(self, m_load, *args, **kwargs):
return super().run(*args, **kwargs)


This doesn't work:

def run(self, *args, **kwargs):
with mock.patch.dict("config.data", {"test": 123}) and mock.patch(
"config.load_data", return_value=None, autospec=True
) as m_load:
return super().run(*args, **kwargs)


The config.load_data is patched so the data doesn't reload, and the module is sort of a singleton.

Continue reading...