How to use pytest.mark.parametrize with fixtures

I want to parametrize a test with fixtures. Something like this:

def count_lines(path):
    with open(path, 'r') as fh:
        return len(fh.readlines())


@pytest.fixture
def data_path():
    return '/path/to/test/data'


@pytest.fixture
def simple_file(data_path):
    return os.path.join(data_path, 'simpletestfile.ext')


@pytest.fixture
def other_file(data_path):
    return os.path.join(data_path, 'othertestfile.ext')


@pytest.mark.parametrize('given, expected', [
    (simple_file, 3),
    (other_file, 9),
])
def test_count_lines(given, expected):
    observed = count_lines(given)
    assert observed == expected

The code above fails because instead of resolving the sample_file and other_file fixtures, the test receives the functions which makes the test fail at the open() statement.

To make it work I have to do something like this:

@pytest.mark.parametrize('given, expected', [
    (simple_file(data_path()), 3),
    (other_file(data_path()), 9),
])

But that seems to defeat the purpose of py.test.

Another alternative is

class TestCountLines(object):

    TESTDATA = [
        {
            'id': 'simple case',
            'input_file': 'simpletestfile.ext',
            'expected': 3,
        },
        {
            'id': 'other case',
            'input_file': 'othertestfile.ext',
            'expected': 9,
        },
    ]

    @pytest.fixture(params=TESTDATA, ids=lambda x: x['id'])
    def staged_testdata(self, request, data_path):
        datafile = os.path.join(data_path, request.param['input_file'])
        expected = request.param['expected']
        yield datafile, expected

    def test_count_lines(self, staged_testdata):
        given, expected = staged_testdata
        observed = count_lines(given)
        assert observed == expected

Yet I find that the code above is curious because the staged_testdata fixture yields two things and then I need to unpack the test data before testing. So, how do I parametrize a test with fixtures?