Adding Tests#
If you are adding new features to manim, you should add appropriate tests for them. Tests prevent manim from breaking at each change by checking that no other feature has been broken and/or been unintentionally modified.
How Manim tests#
Manim uses pytest as its testing framework. To start the testing process, go to the root directory of the project and run pytest in your terminal. Any errors that occur during testing will be displayed in the terminal.
Some useful pytest flags:
-x
will make pytest stop at the first failure it encounters-s
will make pytest display all the print messages (including those during scene generation, like DEBUG messages)--skip_slow
will skip the (arbitrarily) slow tests--show_diff
will show a visual comparison in case a unit test is failing.
How it Works#
At the moment there are three types of tests:
Unit Tests:
Tests for most of the basic functionalities of manim. For example, there a test for
Mobject
, that checks if it can be added to a Scene, etc.Graphical unit tests: Because
manim
is a graphics library, we test frames. To do so, we create test scenes that render a specific feature. When pytest runs, it compares the result of the test to the control data, either at 6 fps or just the last frame. If it matches, the tests pass. If the test and control data differ, the tests fail. You can use--show_diff
flag withpytest
to visually see the differences. Theextract_frames.py
script lets you see all the frames of a test.Videos format tests:
As Manim is a video library, we have to test videos as well. Unfortunately, we cannot directly test video content as rendered videos can differ slightly depending on the system (for reasons related to ffmpeg). Therefore, we only compare video configuration values, exported in .json.
Architecture#
The manim/tests
directory looks like this:
.
├── conftest.py
├── control_data
│ ├── graphical_units_data
│ │ ├── creation
│ │ │ ├── DrawBorderThenFillTest.npy
│ │ │ ├── FadeInFromDownTest.npy
│ │ │ ├── FadeInFromLargeTest.npy
│ │ │ ├── FadeInFromTest.npy
│ │ │ ├── FadeInTest.npy
│ │ │ ├── ...
│ │ ├── geometry
│ │ │ ├── AnnularSectorTest.npy
│ │ │ ├── AnnulusTest.npy
│ │ │ ├── ArcBetweenPointsTest.npy
│ │ │ ├── ArcTest.npy
│ │ │ ├── CircleTest.npy
│ │ │ ├── CoordinatesTest.npy
│ │ │ ├── ...
│ │ ├── graph
│ │ │ ├── ...
| | | | ...
│ └── videos_data
│ ├── SquareToCircleWithDefaultValues.json
│ └── SquareToCircleWithlFlag.json
├── helpers
│ ├── graphical_units.py
│ ├── __init__.py
│ └── video_utils.py
├── __init__.py
├── test_camera.py
├── test_config.py
├── test_copy.py
├── test_vectorized_mobject.py
├── test_graphical_units
│ ├── conftest.py
│ ├── __init__.py
│ ├── test_creation.py
│ ├── test_geometry.py
│ ├── test_graph.py
│ ├── test_indication.py
│ ├── test_movements.py
│ ├── test_threed.py
│ ├── test_transform.py
│ └── test_updaters.py
├── test_logging
│ ├── basic_scenes.py
│ ├── expected.txt
│ ├── testloggingconfig.cfg
│ └── test_logging.py
├── test_scene_rendering
│ ├── conftest.py
│ ├── __init__.py
│ ├── simple_scenes.py
│ ├── standard_config.cfg
│ └── test_cli_flags.py
└── utils
├── commands.py
├── __init__.py
├── testing_utils.py
└── video_tester.py
...
The Main Directories#
control_data/
:The directory containing control data.
control_data/graphical_units_data/
contains the expected and correct frame data for graphical tests, andcontrol_data/videos_data/
contains the .json files used to check videos.test_graphical_units/
:Contains graphical tests.
test_scene_rendering/
:For tests that need to render a scene in some way, such as tests for CLI flags (end-to-end tests).
utils/
:Useful internal functions used by pytest.
Note
fixtures are not contained here, they are in
conftest.py
.helpers/
:Helper functions for developers to setup graphical/video tests.
Adding a New Test#
Unit Tests#
Pytest determines which functions are tests by searching for files whose
names begin with “test_”, and then within those files for functions
beginning with “test” and classes beginning with “Test”. These kinds of
tests must be in tests/
(e.g. tests/test_container.py
).
Graphical Unit Test#
The test must be written in the correct file (i.e. the file that corresponds to the appropriate category the feature belongs to) and follow the structure of unit tests.
For example, to test the Circle
VMobject which resides in
manim/mobject/geometry.py
, add the CircleTest to
test/test_geometry.py
.
The name of the module is indicated by the variable __module_test__, that must be declared in any graphical test file. The module name is used to store the graphical control data.
Important
You will need to use the frames_comparison
decorator to create a test. The test function must accept a
parameter named scene
that will be used like self
in a standard construct
method.
Here’s an example in test_geometry.py
:
from manim import *
from manim.utils.testing.frames_comparison import frames_comparison
__module_test__ = "geometry"
@frames_comparison
def test_circle(scene):
circle = Circle()
scene.play(Animation(circle))
The decorator can be used with or without parentheses. By default, the test only tests the last frame. To enable multi-frame testing, you have to set ``last_frame=False`` in the parameters.
@frames_comparison(last_frame=False)
def test_circle(scene):
circle = Circle()
scene.play(Animation(circle))
You can also specify, when needed, which base scene you need (ThreeDScene, for example) :
@frames_comparison(last_frame=False, base_scene=ThreeDScene)
def test_circle(scene):
circle = Circle()
scene.play(Animation(circle))
Feel free to check the documentation of @frames_comparison
for more.
Note that tests name must follow the syntax test_<thing_to_test>
, otherwise pytest will not recognize it as a test.
Warning
If you run pytest now, you will get a FileNotFound
error. This is because
you have not created control data for your test.
To create the control data for your test, you have to use the flag --set_test
along with pytest.
For the example above, it would be
pytest test_geometry.py::test_circle --set_test -s
(-s
is here to see manim logs, so you can see what’s going on).
If you want to see all the control data frames (e.g. to make sure your test is doing what you want), use the
extract_frames.py
script. The first parameter is the path to a .npz
file and the second parameter is the
directory you want the frames created. The frames will be named frame0.png
, frame1.png
, etc.
python scripts/extract_frames.py tests/test_graphical_units/control_data/plot/axes.npz output
Please make sure to add the control data to git as soon as it is produced with git add <your-control-data.npz>
.
Videos tests#
To test videos generated, we use the decorator
tests.utils.videos_tester.video_comparison
:
@video_comparison(
"SquareToCircleWithlFlag.json", "videos/simple_scenes/480p15/SquareToCircle.mp4"
)
def test_basic_scene_l_flag(tmp_path, manim_cfg_file, simple_scenes_path):
scene_name = "SquareToCircle"
command = [
"python",
"-m",
"manim",
simple_scenes_path,
scene_name,
"-l",
"--media_dir",
str(tmp_path),
]
out, err, exit_code = capture(command)
assert exit_code == 0, err
Note
assert exit*\ code == 0, err
is used in case of the command fails
to run. The decorator takes two arguments: json name and the path
to where the video should be generated, starting from the media/
dir.
Note the fixtures here:
tmp_path is a pytest fixture to get a tmp_path. Manim will output here, according to the flag
--media_dir
.manim_cfg_file
fixture that return a path pointing totest_scene_rendering/standard_config.cfg
. It’s just to shorten the code, in the case multiple tests need to use this cfg file.simple_scenes_path
same as above, except fortest_scene_rendering/simple_scene.py
You have to generate a .json
file first to be able to test your video. To
do that, use helpers.save_control_data_from_video
.
For instance, a test that will check if the l flag works properly will first
require rendering a video using the -l flag from a scene. Then we will test
(in this case, SquareToCircle), that lives in
test_scene_rendering/simple_scene.py
. Change directories to tests/
,
create a file (e.g. create\_data.py
) that you will remove as soon as
you’re done. Then run:
save_control_data_from_video("<path-to-video>", "SquareToCircleWithlFlag.json")
Running this will save
control_data/videos_data/SquareToCircleWithlFlag.json
, which will
look like this:
{
"name": "SquareToCircleWithlFlag",
"config": {
"codec_name": "h264",
"width": 854,
"height": 480,
"avg_frame_rate": "15/1",
"duration": "1.000000",
"nb_frames": "15"
}
}
If you have any questions, please don’t hesitate to ask on Discord, in your pull request, or in an issue.