Mocking function calls in Python - inspired by Ruby's RSpec-Mocks.

Test environments are usually isolated from external services, and meant to check the execution logic of the code exclusively. However it is quite common for projects to deal with external APIs (to receive or post data for example) or systems (such as databases). In that scenario, there are (at least) 2 options:

1. not testing such modules or objects to avoid performing external calls (that would fail anyway - ideally),
2. mocking external calls in order to test their surrounding logic, and increase the coverage of tests as much as possible.

This package is here to help with 2).

This repo contains:

• the expectise module itself, under /expectise;
• a dummy example of small module and its tests under /example with unit tests showcasing what the expectise package can do.

Install from Pypi:

pip install expectise

There are 2 steps in the lifecycle of decoration:

1. Set up: marking a method, so that it can be replaced by a surrogate method, and its calls intercepted;
2. Tear down: resetting the mocking behaviour so that all unit tests are fully independent and don't interfere with each other. During that step, some infractions can be caught too, such as not having called a method that was supposed to be called.

Methods can be marked as mocked in 2 different ways, that are described below.

1. Method 1: using the mock_if decorator, along with the name and value of the environment variable you use to identify your test environment. This environment variable, say ENV will be checked at interpretation time: if its value matches the input, say ENV=test, the mocking logic will be implemented; if not, nothing in your code will be modified, and performance will stay the same since nothing will happen passed interpretation.

Example of decoration:

from expectise import mock_if


class MyObject:
    ...

    @mock_if("ENV", "test")
    def my_function(self, ...)
        ...

    ...

This method is concise, explicit and transparent: you can identify mocked candidates at a glance, and your tests can remain light without any heavy setup logic. However, it means patching production code, and carrying a dependency on this package in your production environment, which may be seen as a deal breaker from an isolation of concerns perspective.

2. Method 2: using explicit mock statements when setting up your tests. Before running individual tests, mocks can be injected explicitly as part of any piece of custom logic, typically through fixtures if you're familiar with pytest (you'll find examples in examples/tests/).

Example of statement:

import pytest
from expectise import mock


@pytest.fixture(autouse=True)
def run_around_tests():
    mock(SomeObject, SomeObject.some_method, "ENV", "test")
    yield
    # see next section for more details on tear down actions

This method is a little bit heavier, and may require more maintenance when mocked objects are modified. However, it keeps a clear separation of concerns with production code that is not patched and does not have to depend on this package.

Once a test has run, underlying expectise objects have to be reset so that 1) some final checks can happen, and 2) new tests can be run with no undesirable side effects from previous tests. Again, there are 2 ways of performing the necessary tear down actions, described below.

1. Method 1: using the Expectations context manager provided in this package. We recommend using this approach if only a few of your tests deal with functions that you want to mock. Toy example:

from expectise import Expect


def test_instance_method():
    with Expectations():
        Expect("SomeAPI").to_receive("update_attribute").and_return("sshhhh")
        ...
        assert SomeAPI().update_attribute("secret_value") == "sshhhh"

2. Method 2: by performing a teardown method for all your tests. We recommend using this approach if most of your tests manipulate objects that you want to mock. Reusing the pytest fixtures example from the previous section:

import pytest
from expectise import mock
from expectise import tear_down


@pytest.fixture(autouse=True)
def run_around_tests():
    mock(SomeObject, SomeObject.some_method, "ENV", "test")
    yield
    tear_down()

Sometimes it can be useful to manually disable a mock - for example, to write a test for a method decorated with mock_if. To achieve this, simply call Expect.disable_mock("<class name>", "<method name>")

The following use cases are covered:

• asserting that a method is called (the right number of times),
• checking the arguments passed to a method,
• overriding a method so that it returns a given output when called,
• overriding a method so that it raises an exception when called.

The above features can be combined too, with the following 4 possible patterns:

Expect('MyObject').to_receive('my_method').and_return(my_object)
Expect('MyObject').to_receive('my_method').and_raise(my_error)
Expect('MyObject').to_receive('my_method').with_args(*my_args, **my_kwargs).and_return(my_object)
Expect('MyObject').to_receive('my_method').with_args(*my_args, **my_kwargs).and_raise(my_error)

A given method of a class can be decorated several times, with different arguments to check and ouputs to be returned. You just have to specify it with several Expect statements. In this case, the order of the statements matters.

The following is valid and assumes my_method is going to be called three times exactly:

Expect('MyObject').to_receive('my_method').with_args(*my_args_1, **my_kwargs_1).and_return(my_object_1)
Expect('MyObject').to_receive('my_method').with_args(*my_args_2, **my_kwargs_2).and_raise(my_error)
Expect('MyObject').to_receive('my_method').with_args(*my_args_3, **my_kwargs_3).and_return(my_object_2)

Note that if a method decorated at least once with an Expect statement is called more or less times than the number of Expect statements, the unit test will fail.

We recommend using asdf for managing high level dependencies. With asdf installed,

1. simply run asdf install at the root of the repository,
2. run poetry install to install python dependencies.

poetry shell
ENV=test python -m pytest -v examples/tests/