views:

39

answers:

2

Hi,

I'm testing a function that obtains a skeleton object from one helper object, modifies it using a second helper, and passes the modified object back to the first helper. Something along the lines of:

class ReadModifyUpdate(object):
    def __init__(self, store, modifier):
        self._store = store
        self._modifier = modifier

    def modify(key):
        record = self._store.read(key)
        self._modifier.modify(record)
        self._store.update(key, record)

Using Python and Mox, we can test this with:

class ReadModifyUpdateTest(mox.MoxTestBase):
    def test_modify(self):
        mock_record = self.mox.CreateMockAnthing()
        mock_store = self.mox.CreateMockAnything()
        mock_modifier = self.mox.CreateMockAnything()

        mock_store.read("test_key").AndReturn(mock_record)
        mock_modifier.modify(mock_record)
        mock_store.update("test_key", mock_record)
        self.mox.ReplayAll()

        updater = ReadModifyUpdate(mock_store, mock_modifier)
        updater.modify("test_key")

...but this doesn't catch the bug in which store.update() is inadvertently called before modifier.modify(). Is there a good way, in Mox, to check the order of methods called on multiple mocks? Something like EasyMock's MocksControl object?

A: 

Maybe not the best solution but you could try to use one mock that you give twice to your object under test. You then have control over the call order.

class ReadModifyUpdateTest(mox.MoxTestBase):
    def test_modify(self):
        mock_record = self.mox.CreateMockAnthing()
        mock_storeModifier = self.mox.CreateMockAnything()

        mock_storeModifier.read("test_key").AndReturn(mock_record)
        mock_storeModifier.modify(mock_record)
        mock_storeModifier.update("test_key", mock_record)
        self.mox.ReplayAll()

        updater = ReadModifyUpdate(mock_storeModifier, mock_storeModifier)
        updater.modify("test_key")
Rod
Thanks for the response, and an interesting suggestion. Using one mock to cover two objects feels strange! The obvious drawback is that it doesn't catch bugs in which we call the wrong methods on the wrong objects - then again, these bugs are probably unlikely than ordering problems in this particular scenario.
Ian Parkinson
A: 

To provide an answer to my own question - I've currently got this working using a side effect which checks the call order.

Defining a helper class:

class OrderedCallSequence(object):
    def __init__(self, test_case):
        self._expectation_count = 0
        self._evaluated = 0
        self._test_case = test_case

    def assertOrder(self):
        self._expectation_count += 1
        expected_position = self._expectation_count

        def side_effect(*args, **kwargs):
            self._evaluated += 1
            self._test_case.assertEquals(self._evaluated, expected_position,
                                         msg="Invoked in incorrect sequence")
        return side_effect

...the test case becomes:

class ReadModifyUpdateTest(mox.MoxTestBase):
    def test_modify(self):
        mock_record = self.mox.CreateMockAnthing()
        mock_store = self.mox.CreateMockAnything()
        mock_modifier = self.mox.CreateMockAnything()

        sequence = OrderedCallSequence(self)
        mock_store.read("test_key").WithSideEffects(sequence.assertOrder()).AndReturn(mock_record)
        mock_modifier.modify(mock_record).WithSideEffects(sequence.assertOrder())
        mock_store.update("test_key", mock_record).WithSideEffects(sequence.assertOrder())
        self.mox.ReplayAll()

        updater = ReadModifyUpdate(mock_store, mock_modifier)
        updater.modify("test_key")
Ian Parkinson