FIX: Guard against already-removed labels in ContourSet.remove() by buddy0452004 · Pull Request #31431 · matplotlib/matplotlib

buddy0452004 · 2026-04-01T06:51:11Z

When contour labels are manually removed before calling CS.remove(), matplotlib crashes with a ValueError.

Based on discussion in #31404, this PR:

Adds a UserWarning in ContourSet.remove() when labels were already manually removed, guiding users to remove the entire ContourSet and recreate it instead.
Adds a note in clabel()'s Returns documentation that the returned Text instances should not be individually removed or have their geometry modified.
Adds a regression test that expects the UserWarning.

Closes #31404

AI Disclosure

No AI used.

PR Checklist

"closes [Bug]: Crash when removing contour set after removing contour labels #31404" is in the body of the PR description
new and changed code is tested
[N/A] Plotting related features are demonstrated in an example
[N/A] New Features and API Changes are noted
Documentation complies with guidelines

rcomer

Thanks @buddy0452004, but I do not think you understood the root cause of why the failure happens. I think the explanation at #31404 (comment) is correct. The error shown in the issue states that we tried to remove something from a list that is not in the list. So we got hold of the list axes.texts without problem, but the labels were no longer in there.

rcomer · 2026-04-01T07:30:08Z

lib/matplotlib/tests/test_contour.py

+
+def test_contour_remove_after_label_removed():
+    # Test that CS.remove() works even if labels were manually removed first
+    # Regression test for https://github.com/matplotlib/matplotlib/issues/31404


We do not typically reference issues in test comments.

rcomer · 2026-04-01T07:34:25Z

lib/matplotlib/tests/test_contour.py

+    x = np.linspace(-3.0, 3.0, 20)
+    y = np.linspace(-2.0, 2.0, 20)
+    X, Y = np.meshgrid(x, y)
+    Z = np.exp(-X**2 - Y**2)


For efficiency, try to use the simplest data input you can that still makes the test effective. Look at test_contourf_rasterize for an example.

@rcomer Yes, that makes sense , I had also suggested using simpler input data in the discussion #31404 .

rcomer · 2026-04-01T07:36:48Z

lib/matplotlib/tests/test_contour.py

+
+
+
+def test_contour_remove_after_label_removed():


Since there is already a test_contour_remove further up the file, I would put this one right below it, to keep tests with similar purposes together.

rcomer · 2026-04-01T07:57:33Z

lib/matplotlib/contour.py

-        for text in self.labelTexts:
-            text.remove()
+        for text in list(self.labelTexts):
+            if axes is not None and text in axes.texts:


Why do we check whether axes is None here?

buddy0452004 · 2026-04-01T11:22:01Z

@rcomer Thanks for the review! You're right about the root cause the issue is that the labels were already removed from axes.texts, not that axes is None. Updated the fix to use try/except ValueError instead, which is simpler and correctly addresses the actual failure.

Also:

Removed the issue reference from the test comment
Simplified test data to use np.arange(16).reshape((4, 4))
Moved the test right below test_contour_remove

timhoffm · 2026-04-01T12:31:17Z

I'm putting this on hold until we have clarified what is actually needed. #31404 (comment)

buddy0452004 · 2026-04-03T06:24:00Z

@timhoffm @rcomer Updated based on the discussion on #31404:

Added a UserWarning in the except block explaining that some labels were manually removed and guiding users to remove the entire ContourSet and recreate it instead.
Added a note in clabel()'s Returns documentation that the returned Text instances should not be individually removed or have their geometry modified.

Chirag3841 · 2026-04-03T07:18:56Z

@timhoffm @rcomer Updated based on the discussion on #31404:

Added a UserWarning in the except block explaining that some labels were manually removed and guiding users to remove the entire ContourSet and recreate it instead.

Added a note in clabel()'s Returns documentation that the returned Text instances should not be individually removed or have their geometry modified.

According to me, the minimal approach related to this issue by adding the note in clabel() is enough as these changes affecting the colorbar ticks .

rcomer · 2026-04-03T08:44:38Z

these changes affecting the colorbar ticks .

Sorry, I'm not following. What is happening to the colorbar ticks?

Chirag3841 · 2026-04-03T09:08:33Z

these changes affecting the colorbar ticks .

Sorry, I'm not following. What is happening to the colorbar ticks?

[gw1] linux -- Python 3.11.15 /opt/hostedtoolcache/Python/3.11.15/x64/bin/python

fmt = '%4.2e'

    @pytest.mark.parametrize('fmt', ['%4.2e', '{x:.2e}'])
    def test_colorbar_format(fmt):
        # make sure that format is passed properly
        x, y = np.ogrid[-4:4:31j, -4:4:31j]
        z = 120000*np.exp(-x**2 - y**2)
    
        fig, ax = plt.subplots()
        im = ax.imshow(z)
        cbar = fig.colorbar(im, format=fmt)
        fig.canvas.draw()
        assert cbar.ax.yaxis.get_ticklabels()[4].get_text() == '8.00e+04'
    
        # make sure that if we change the clim of the mappable that the
        # formatting is *not* lost:
        im.set_clim([4, 200])
        fig.canvas.draw()
        assert cbar.ax.yaxis.get_ticklabels()[4].get_text() == '2.00e+02'
    
        # but if we change the norm:
        im.set_norm(LogNorm(vmin=0.1, vmax=10))
        fig.canvas.draw()
>       assert (cbar.ax.yaxis.get_ticklabels()[0].get_text() ==
                '$\\mathdefault{10^{-2}}$')
E       AssertionError: assert '$\\mathdefault{10^{-1}}$' == '$\\mathdefault{10^{-2}}$'
E         - $\mathdefault{10^{-2}}$
E         ?                    ^
E         + $\mathdefault{10^{-1}}$
E         ?                    ^

lib/matplotlib/tests/test_colorbar.py:636: AssertionError

But I am not 100% sure that, this error is related to code or not .

rcomer · 2026-04-03T09:15:42Z

That one is a sporadic flaky test, so not related to this PR.

lib/matplotlib/contour.py

lib/matplotlib/tests/test_contour.py

buddy0452004 · 2026-04-03T12:20:32Z

I had removed all labels in a loop thinking it would better trigger the warning, and added a second cs.clabel() call thinking it would make the scenario more complete but neither was relevant to what the test is actually checking. Simplified to just removing one label as suggested.
@timhoffm all your review feedback has been addressed removed the list() wrapping, switched to _api.warn_external() fixed the docstring wording and backticks, and simplified the test. Thanks for the detailed review.

lib/matplotlib/contour.py

Co-authored-by: Tim Hoffmann <[email protected]>

FIX: Guard against already removed labels in ContourSet.remove()

d157587

github-actions bot added the topic: contour label Apr 1, 2026

FIX: Add blank lines before test function

ada7b70

buddy0452004 mentioned this pull request Apr 1, 2026

FIX: Guard against already-removed labels in ContourSet.remove() #31427

Closed

1 task

rcomer reviewed Apr 1, 2026

View reviewed changes

FIX: Address review feedback - simpler fix and cleaner test

ff2eab9

FIX: Remove trailing newline

8038b61

timhoffm marked this pull request as draft April 1, 2026 12:31

FIX: Add warning and docs for manual label removal in ContourSet

d35b9cd

TST: Expect UserWarning when labels manually removed before CS.remove()

ca0515a

timhoffm reviewed Apr 3, 2026

View reviewed changes

buddy0452004 added 2 commits April 3, 2026 17:13

FIX: Address timhoffm review feedback

b23fd27

FIX: Fix line too long in docstring

8eff292

timhoffm approved these changes Apr 3, 2026

View reviewed changes

timhoffm marked this pull request as ready for review April 3, 2026 13:50

timhoffm reviewed Apr 3, 2026

View reviewed changes

lib/matplotlib/contour.py Outdated Show resolved Hide resolved

Update lib/matplotlib/contour.py

5f1d6b0

Co-authored-by: Tim Hoffmann <[email protected]>

Uh oh!

Conversation

buddy0452004 commented Apr 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

AI Disclosure

PR Checklist

Uh oh!

rcomer left a comment

Choose a reason for hiding this comment

Uh oh!

rcomer Apr 1, 2026

Choose a reason for hiding this comment

Uh oh!

rcomer Apr 1, 2026

Choose a reason for hiding this comment

Uh oh!

Chirag3841 Apr 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rcomer Apr 1, 2026

Choose a reason for hiding this comment

Uh oh!

rcomer Apr 1, 2026

Choose a reason for hiding this comment

Uh oh!

buddy0452004 commented Apr 1, 2026

Uh oh!

timhoffm commented Apr 1, 2026

Uh oh!

buddy0452004 commented Apr 3, 2026

Uh oh!

Chirag3841 commented Apr 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rcomer commented Apr 3, 2026

Uh oh!

Chirag3841 commented Apr 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rcomer commented Apr 3, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

buddy0452004 commented Apr 3, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

buddy0452004 commented Apr 1, 2026 •

edited

Loading

Chirag3841 Apr 1, 2026 •

edited

Loading

Chirag3841 commented Apr 3, 2026 •

edited

Loading

Chirag3841 commented Apr 3, 2026 •

edited

Loading