Python pytest mock fails with "assert None" for function call assertions

Question

I am trying to mock some calls to boto3 and it looks like the mocked function is returning the correct value, and it look like if I change the assertion so it no longer matches what was passed in the assertion fails because the input parameters do not match, however if I make them match then the assertion fails with:

E       AssertionError: assert None
E        +  where None = <bound method wrap_assert_called_with of <MagicMock name='get_item' id='139668501580240'>>(TableName='TEST_TABLE', Key={'ServiceName': {'S': 'Site'}})
E        +    where <bound method wrap_assert_called_with of <MagicMock name='get_item' id='139668501580240'>> = <MagicMock name='get_item' id='139668501580240'>.assert_called_with
E        +      where <MagicMock name='get_item' id='139668501580240'> = <botocore.client.DynamoDB object at 0x7f071b8251f0>.get_item
E        +        where <botocore.client.DynamoDB object at 0x7f071b8251f0> = site_dao.ddb_client

The dynamo DB object is a global variable.

ddb_client = boto3.client("dynamodb")

def query_ddb():
        """Query dynamo DB for the current strategy.

        Returns:
            strategy (str): The latest strategy from dynamo DB.

        """
        response = None

        try:
            ddb_table = os.environ["DDB_DA_STRATEGY"]

            response = ddb_client.get_item(
                TableName=ddb_table, Key={"ServiceName": {"S": "Site"}}
            )
        except Exception as exception:
            LOGGER.error(exception)

            raise ServerException("The server was unable to process your request.", [])

        return response.get("Item").get("Strategy").get("S")

And my unit test looks like:

def test_data_access_context_strategy_ddb(mocker):
    object_key = {
        "ServiceName": {"S": "Site"}
    }

    table_name = "TEST_TABLE"

    expected_response = "SqlServer"

    os_env = {
        "DDB_DA_STRATEGY": table_name,
    }

    ddb_response = {
        "Item": {
            "Strategy": {
                "S": expected_response
            }
        }
    }

    mocker.patch.dict(os.environ, os_env)

    mocker.patch.object(site_dao.ddb_client, "get_item")

    site_dao.ddb_client.get_item.return_value = ddb_response

    data_access_context = site_dao.DataAccessContext()

    response = data_access_context.query_ddb()
    
    assert response == expected_response

    assert site_dao.ddb_client.get_item.assert_called_with(TableName=table_name, Key=object_key)

I cant work out what is going wrong, if I change the expected value for assert_called_with, so for example:

assert site_dao.ddb_client.get_item.assert_called_with(TableName="TT", Key=object_key)

The test fails with:

E       AssertionError: expected call not found.
E       Expected: get_item(TableName='TT', Key={'ServiceName': {'S': 'Site'}})
E       Actual: get_item(TableName='TEST_TABLE', Key={'ServiceName': {'S': 'Site'}})
E       
E       pytest introspection follows:
E       
E       Kwargs:
E       assert {'Key': {'Ser... 'TEST_TABLE'} == {'Key': {'Ser...leName': 'TT'}
E         Omitting 1 identical items, use -vv to show
E         Differing items:
E         {'TableName': 'TEST_TABLE'} != {'TableName': 'TT'}
E         Use -v to get the full diff

So when the expected and actual inputs differ then it fails because of this, however when they are the same and the test should pass it fails because it is as if the function were never even called.

Answer 1

You have two assertions on this line:

assert site_dao.ddb_client.get_item.assert_called_with(TableName=...)

The first assertion is assert_called_with which sounds like it is what you want. Then you have another assertion at the beginning of the line: assert ... which asserts on the return value of the assert_called_with function. That return value is None when the assertion passes. So then the whole line evaluates to assert None, and None is false-y, so you end up with assert False.

tl;dr Don't use assert twice.