pytest-data documentation


pip install pytest-data


Useful functions for managing data for pytest fixtures.

pytest_data.functions.get_data(request, attribute_name, default_data={})[source]

Returns merged data from module, class and function. The most highest priority have data nearest to test code. It means that data on function have top priority, then class, then module, then param of request and at last default_data.

Example of fixture:

def client(request):
    client_data = get_data(request, 'client_data', {'name': 'Jerry', 'address': 'somewhere'})
    return Client(client_data)

And example how to use data in your test:

client_data = {'name': 'Sheldon', 'age': 30}

class ClientTest:
    client_data = {'iq': 200, 'foo': 'bar'}

    @use_data(client_data={'foo': 'baz'})
    def test_foo(self, client):
        assert == 'Sheldon'
        assert client.address == 'somewhere'
        assert client.age == 30
        assert == 200
        assert == 'baz'

Most of the time you will need only dictionary but sometimes is needed list of dictionaries. It is supported as well but you need to know how it works. It will find longest list and rest of them will just cycle around.

def clients(request):
    clients_data = get_data(request, 'client_data', {'name': 'Jerry'})
    return [Client(client_data) for client_data in clients_data]

client_data = [{'age': 20}, {'age': 30}, {'age': 40}]

@use_data(client_data=[{'foo': 'bar', 'foo': 'baz'}])
def test_foo(self, clients):
    assert clients[0].name == clients[1].name == clients[2].name
    assert clients[0].age == 20
    assert clients[0].foo == 'bar'
    assert clients[1].age == 30
    assert clients[1].foo == 'baz'
    assert clients[2].age == 40
    assert clients[2].foo == 'bar'

It also works with parametrized fixture. Following fixture will then call each test twice for two different types of user and you are still able to modify some data of user when needed:

@pytest.fixture(params=[{'registred': True}, {'registred': False}])
def user(request):
    user_data = get_data(request, 'user_data', {'name': 'Jerry'})
    return User(user_data)

Decorator make sexier assigning of fixture data on function. Instead of writing this code:

def test_foo():
test_foo.client_data = {...}

you can write this one. It’s good that used data are visible on top of test function and not somewhere at the end.

def test_foo():

Changed in version 0.3: use_data can be used only for tests, not fixtures. So since version 0.3 there is check that it’s used only with tests (function starts with test).


use_data() mixed with pytest’s parametrize. The best way how to describe it is to show some code:

@use_data_parametrize(client=[{...}, {...}], user=[{...}, {...}])
def test_foo(client, user):
    assert 0

def client(request):
    return get_data(request, 'client_data', {...})

def user(request):
    return get_data(request, 'user_data', {...})

And it will call test test_foo four times, for every combination of client and user. As you would expect by pytest’s mark parametrize.

========= FAILURES ==========
-- test_foo[client0-user0] --
-- test_foo[client0-user1] --
-- test_foo[client1-user0] --
-- test_foo[client1-user1] --

Just note that in use_data() is used as key attribute_name defined in fixture by calling get_data() whereas here is used fixture name. Because pytest need to assign params to fixtures. Data passed to this method is then found in fixture’s request.param instead of request.function.attribute_name.

New in version 0.2.