Reorganise test fixtures #380
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #380 +/- ##
=======================================
Coverage 99.79% 99.79%
=======================================
Files 14 14
Lines 969 969
=======================================
Hits 967 967
Misses 2 2 ☔ View full report in Codecov by Sentry. |
lochhh
force-pushed
the
reorg-fixtures
branch
from
cc622b9 to
978c8b3
Compare
lochhh
force-pushed
the
reorg-fixtures
branch
from
7c0b12c to
7acf596
Compare
lochhh
force-pushed
the
reorg-fixtures
branch
from
60b3e50 to
253a8ea
Compare
There was a problem hiding this comment.
thanks @lochhh it looks fantastic!
I really like the consistency (such as the _file endings, the via_ prefixes, with_nan suffixes etc), it makes a lot of it much more readable. I followed some of these ideas and suggested a few invalid_ prefixes for some dataset-related fixtures.
The only larger thing is that I had a go at moving the movement_dataset_asserts fixture to be under the helpers fixtures, since the class that wrapped it only had that one method - it's here, but you can revert if you disagree.
Re PR #385, since that is now merged, if you want to deal with that in this PR I am happy to review any additions. Otherwise we can deal with it separately in another PR/issue.
| * expected_n_nans_in_position | ||
| @pytest.mark.parametrize( | ||
| "window", | ||
| [3, 5, 6, 10], # data is nframes = 10 |
There was a problem hiding this comment.
| [3, 5, 6, 10], # data is nframes = 10 | |
| [3, 5, 6, 10], # input data has 10 frames |
| - id_1 moves along x=-y line from the origin | ||
| - they both move one unit (pixel) along each axis in each frame | ||
| class TestComputeKinematics: | ||
| """Test ``compute_[kinematic_variable]`` with valid and invalid inputs.""" |
There was a problem hiding this comment.
| """Test ``compute_[kinematic_variable]`` with valid and invalid inputs.""" | |
| """Test ``compute_[kinematic_variable]`` with valid and invalid inputs. | |
| When valid, the expected kinematics are a uniform linear motion case. | |
| Uniform linear motion means the individuals move along a line at constant | |
| velocity. | |
| We consider 2 individuals ("id_0" and "id_1"), tracked for 10 frames, | |
| along x and y: | |
| - Individual 0 ("id_0") moves along x=y line from the origin | |
| - Individual 1 ("id_1") moves along x=-y line from the origin | |
| - they both move one unit (pixel) along each axis in each frame | |
| """ |
I think this description may have got lost somewhere
| """Test kinematics computation with an invalid dataset.""" | ||
| with expected_exception: | ||
| position = request.getfixturevalue(invalid_dataset).position | ||
| getattr(kinematics, f"compute_{kinematic_variable}")(position) | ||
|
|
||
|
|
||
| @pytest.mark.parametrize("order", [0, -1, 1.0, "1"]) |
There was a problem hiding this comment.
Below a suggestion I realise that is out of scope for this PR - if you want it can be dealt separately. It involves:
- renaming the test (it refers to the old name of the function I think)
- include the expected exception in the parametrisation for readability
@pytest.mark.parametrize(
"order, expected_exception",
[
(0, pytest.raises(ValueError)),
(-1, pytest.raises(ValueError)),
(1.0, pytest.raises(TypeError)),
("1", pytest.raises(TypeError)),
],
)
def test_time_derivative_with_invalid_order(order, expected_exception):
"""Test that an error is raised when the order is non-positive."""
data = np.arange(10)
with expected_exception:
kinematics.compute_time_derivative(data, order=order)
Oops sorry, we merged it without looking here. |
|
Description
What is this PR
Why is this PR needed?
This PR closes #336 and #222.
What does this PR do?
This PR
data.py,files.py, orhelpers.pyintests/fixtures/; where data contains fixtures for valid and invalid movement datasets and arrays, dataframes, and themovement_dataset_asserts.filefixtures returning file paths are suffixed with "_file" to enable devs to easily identify file fixtures and search for existing onesvia_valid_poses_dataset,valid_poses_dataset_with_nan,valid_position_arrayfixtures with the neweruniform_linear_motionfixtures and updates the affected testsposes_with_nanfixture to match that of bboxes (i.e. both centroids have nans at the same time points)datafixtures containing nans are consistently suffixed with_with_nan(before we had a mix of_with_nansand_with_nan)TestComputeKinematics,TestFilteringValidDatasetid_0(before, poses ids start fromid_1); semi-related note: movement by default sets "individuals" names to begin with "individual_0" if none are provided upon import.I haven't changed the vector-related tests in
test_kinematics.pyas there is ongoing work in PR #385.References
#336 #222
How has this PR been tested?
Tests passing
Checklist: