matplotlib.testing#

matplotlib.testing#

Helper functions for testing.

matplotlib.testing.ipython_in_subprocess(requested_backend_or_gui_framework, all_expected_backends)[source]#

matplotlib.testing.is_ci_environment()[source]#

matplotlib.testing.set_font_settings_for_testing()[source]#

matplotlib.testing.set_reproducibility_for_testing()[source]#

matplotlib.testing.setup()[source]#

matplotlib.testing.subprocess_run_for_testing(command, env=None, timeout=60, stdout=None, stderr=None, check=False, text=True, capture_output=False, **kwargs)[source]#

Create and run a subprocess.

Thin wrapper around subprocess.run, intended for testing. Will mark fork() failures on Cygwin as expected failures: not a success, but not indicating a problem with the code either.

Parameters:

argslist of str

envdict[str, str]

timeoutfloat

stdout, stderr

checkbool

textbool

Also called universal_newlines in subprocess. I chose this name since the main effect is returning bytes (False) vs. str (True), though it also tries to normalize newlines across platforms.

capture_outputbool

Set stdout and stderr to subprocess.PIPE

Returns:

procsubprocess.Popen

Raises:

pytest.skip

If running on emscripten, which does not support subprocesses.

pytest.xfail

If platform is Cygwin and subprocess reports a fork() failure.

See also

subprocess.run

matplotlib.testing.subprocess_run_helper(func, *args, timeout, extra_env=None)[source]#

Run a function in a sub-process.

Parameters:

funcfunction

The function to be run. It must be in a module that is importable.

*argsstr

Any additional command line arguments to be passed in the first argument to subprocess.run.

extra_envdict[str, str]

Any additional environment variables to be set for the subprocess.

matplotlib.testing.conftest.text_placeholders(monkeypatch)#

Replace texts with placeholder rectangles.

The rectangle size only depends on the font size and the number of characters. It is thus insensitive to font properties and rendering details. This should be used for tests that depend on text geometries but not the actual text rendering, e.g. layout tests.

matplotlib.testing.compare#

Utilities for comparing image results.

matplotlib.testing.compare.calculate_rms(expected_image, actual_image)[source]#

Calculate the per-pixel errors, then compute the root mean square error.

matplotlib.testing.compare.comparable_formats()[source]#

Return the list of file formats that compare_images can compare on this system.

Returns:

list of str

E.g. ['png', 'pdf', 'svg', 'eps'].

matplotlib.testing.compare.compare_images(expected, actual, tol, in_decorator=False)[source]#

Compare two "image" files checking differences within a tolerance.

The two given filenames may point to files which are convertible to PNG via the converter dictionary. The underlying RMS is calculated in a similar way to the calculate_rms function.

Parameters:

expectedstr

The filename of the expected image.

actualstr

The filename of the actual image.

tolfloat

The tolerance (a color value difference, where 255 is the maximal difference). The test fails if the average pixel difference is greater than this value.

in_decoratorbool

Determines the output format. If called from image_comparison decorator, this should be True. (default=False)

Returns:

None or dict or str

Return None if the images are equal within the given tolerance.

If the images differ, the return value depends on in_decorator. If in_decorator is true, a dict with the following entries is returned:

• rms: The RMS of the image difference.

• expected: The filename of the expected image.

• actual: The filename of the actual image.

• diff_image: The filename of the difference image.

• tol: The comparison tolerance.

Otherwise, a human-readable multi-line string representation of this information is returned.

Examples

img1 = "./baseline/plot.png"
img2 = "./output/plot.png"
compare_images(img1, img2, 0.001)

matplotlib.testing.decorators#

matplotlib.testing.decorators.check_figures_equal(*, extensions=('png',), tol=0)[source]#

Decorator for test cases that generate and compare two figures.

The decorated function must take two keyword arguments, fig_test and fig_ref, and draw the test and reference images on them. After the function returns, the figures are saved and compared.

This decorator should be preferred over image_comparison when possible in order to keep the size of the test suite from ballooning.

Parameters:

extensionslist, default: ["png"]

The extensions to test. Supported extensions are "png", "pdf", "svg".

Testing with the one default extension is sufficient if the output is not format dependent, e.g. if you test that a bar() plot yields the same result as some manually placed Rectangles. You should use all extensions if a renderer property is involved, e.g. correct alpha blending.

tolfloat

The RMS threshold above which the test is considered failed.

Raises:

RuntimeError

If any new figures are created (and not subsequently closed) inside the test function.

Examples

Check that calling Axes.plot with a single argument plots it against [0, 1, 2, ...]:

@check_figures_equal()
def test_plot(fig_test, fig_ref):
    fig_test.subplots().plot([1, 3, 5])
    fig_ref.subplots().plot([0, 1, 2], [1, 3, 5])

matplotlib.testing.decorators.image_comparison(baseline_images, extensions=None, tol=0, freetype_version=None, remove_text=False, savefig_kwarg=None, style=None)[source]#

Compare images generated by the test with those specified in baseline_images, which must correspond, else an ImageComparisonFailure exception will be raised.

Parameters:

baseline_imageslist or None

A list of strings specifying the names of the images generated by calls to Figure.savefig.

If None, the test function must use the baseline_images fixture, either as a parameter or with pytest.mark.usefixtures. This value is only allowed when using pytest.

extensionsNone or list of str

The list of extensions to test, e.g. ['png', 'pdf'].

If None, defaults to: png, pdf, and svg.

When testing a single extension, it can be directly included in the names passed to baseline_images. In that case, extensions must not be set.

In order to keep the size of the test suite from ballooning, we only include the svg or pdf outputs if the test is explicitly exercising a feature dependent on that backend (see also the check_figures_equal decorator for that purpose).

tolfloat, default: 0

The RMS threshold above which the test is considered failed.

Due to expected small differences in floating-point calculations, on 32-bit systems an additional 0.06 is added to this threshold.

freetype_versionstr or tuple

The expected freetype version or range of versions for this test to pass.

remove_textbool

Remove the title and tick text from the figure before comparison. This is useful to make the baseline images independent of variations in text rendering between different versions of FreeType.

This does not remove other, more deliberate, text, such as legends and annotations.

savefig_kwargdict

Optional arguments that are passed to the savefig method.

stylestr, dict, or list

The style(s) to apply to the image test. The test itself can also apply additional styles if desired.

Changed in version 3.11: This defaults to ['classic', '_classic_test_patch'], but will be changing to 'mpl20' as of Matplotlib 3.13. A warning is raised if not explicitly passed.

matplotlib.testing.decorators.remove_ticks_and_titles(figure)[source]#

matplotlib.testing.exceptions#

exception matplotlib.testing.exceptions.ImageComparisonFailure[source]#

Bases: AssertionError

Raise this exception to mark a test as a comparison between two images.

Testing with optional dependencies#

For more information on fixtures, see pytest fixtures.

matplotlib.testing.conftest.pd()#

Fixture to import and configure pandas. Using this fixture, the test is skipped when pandas is not installed. Use this fixture instead of importing pandas in test files.

Examples

Request the pandas fixture by passing in pd as an argument to the test

def test_matshow_pandas(pd):

    df = pd.DataFrame({'x':[1,2,3], 'y':[4,5,6]})
    im = plt.figure().subplots().matshow(df)
    np.testing.assert_array_equal(im.get_array(), df)

matplotlib.testing.conftest.xr()#

Fixture to import xarray so that the test is skipped when xarray is not installed. Use this fixture instead of importing xrray in test files.

Examples

Request the xarray fixture by passing in xr as an argument to the test

def test_imshow_xarray(xr):

    ds = xr.DataArray(np.random.randn(2, 3))
    im = plt.figure().subplots().imshow(ds)
    np.testing.assert_array_equal(im.get_array(), ds)