feat(plotting): support boolean color mappings from adata.uns

[daveringelberg]

This is a draft PR addressing #4219
Release notes are included in docs/release-notes/4219.feat.md

Summary

This draft PR adds support for mappings stored in adata.uns[f"{key}_colors"], so boolean colors can stay tied to True/False after subsetting.

Existing list-based color behavior is preserved.

Testing

Added tests for:

• mapping palettes stored in adata.uns
• boolean mappings after subsetting to only True or only False
• boolean mappings passed through the explicit palette= argument
• regression test for Support dictionaries for Boolean colors in 'adata.uns' #4219 covering boolean color mappings in adata.uns after subsetting.

Local checks run:

• ruff check
• ruff format --check
• python -m py_compile
• git diff --check

I could not run pytest locally due to local environment issues, so I am opening this as a draft for CI and maintainer feedback.

[codecov]

❌ 1 Tests Failed:

Tests completed	Failed	Passed	Skipped
2023	1	2022	1067

View the top 1 failed test(s) by shortest run time

tests/plotting/legacy/test_plotting_utils.py::test_get_palette_bool_mapping_from_palette_argument

Stack Traces | 0.021s run time

#x1B[0m#x1B[94mdef#x1B[39;49;00m#x1B[90m #x1B[39;49;00m#x1B[92mtest_get_palette_bool_mapping_from_palette_argument#x1B[39;49;00m():#x1B[90m#x1B[39;49;00m
        adata = AnnData(#x1B[90m#x1B[39;49;00m
            X=np.ones((#x1B[94m4#x1B[39;49;00m, #x1B[94m1#x1B[39;49;00m)),#x1B[90m#x1B[39;49;00m
            obs={#x1B[33m"#x1B[39;49;00m#x1B[33mflag#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m: [#x1B[94mFalse#x1B[39;49;00m, #x1B[94mTrue#x1B[39;49;00m, #x1B[94mFalse#x1B[39;49;00m, #x1B[94mTrue#x1B[39;49;00m]},#x1B[90m#x1B[39;49;00m
        )#x1B[90m#x1B[39;49;00m
    #x1B[90m#x1B[39;49;00m
>       #x1B[94massert#x1B[39;49;00m _get_palette(adata, #x1B[33m"#x1B[39;49;00m#x1B[33mflag#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m, palette={#x1B[94mFalse#x1B[39;49;00m: #x1B[33m"#x1B[39;49;00m#x1B[33mred#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m, #x1B[94mTrue#x1B[39;49;00m: #x1B[33m"#x1B[39;49;00m#x1B[33mblue#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m}) == {#x1B[90m#x1B[39;49;00m
            #x1B[33m"#x1B[39;49;00m#x1B[33mFalse#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m: #x1B[33m"#x1B[39;49;00m#x1B[33m#ff0000#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m,#x1B[90m#x1B[39;49;00m
            #x1B[33m"#x1B[39;49;00m#x1B[33mTrue#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m: #x1B[33m"#x1B[39;49;00m#x1B[33m#0000ff#x1B[39;49;00m#x1B[33m"#x1B[39;49;00m,#x1B[90m#x1B[39;49;00m
        }#x1B[90m#x1B[39;49;00m
#x1B[1m#x1B[31mE       AssertionError: assert {'False': '#f...: '#0000ffff'} == {'False': '#f...e': '#0000ff'}#x1B[0m
#x1B[1m#x1B[31mE         #x1B[0m
#x1B[1m#x1B[31mE         Differing items:#x1B[0m
#x1B[1m#x1B[31mE         #x1B[0m{#x1B[33m'#x1B[39;49;00m#x1B[33mTrue#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m: #x1B[33m'#x1B[39;49;00m#x1B[33m#0000ffff#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m}#x1B[90m#x1B[39;49;00m != #x1B[0m{#x1B[33m'#x1B[39;49;00m#x1B[33mTrue#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m: #x1B[33m'#x1B[39;49;00m#x1B[33m#0000ff#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m}#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         #x1B[0m{#x1B[33m'#x1B[39;49;00m#x1B[33mFalse#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m: #x1B[33m'#x1B[39;49;00m#x1B[33m#ff0000ff#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m}#x1B[90m#x1B[39;49;00m != #x1B[0m{#x1B[33m'#x1B[39;49;00m#x1B[33mFalse#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m: #x1B[33m'#x1B[39;49;00m#x1B[33m#ff0000#x1B[39;49;00m#x1B[33m'#x1B[39;49;00m}#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         #x1B[0m
#x1B[1m#x1B[31mE         Full diff:#x1B[0m
#x1B[1m#x1B[31mE         #x1B[0m#x1B[90m #x1B[39;49;00m {#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         #x1B[91m-     'False': '#ff0000',#x1B[39;49;00m#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         #x1B[92m+     'False': '#ff0000ff',#x1B[39;49;00m#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         ?                      ++#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         #x1B[91m-     'True': '#0000ff',#x1B[39;49;00m#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         #x1B[92m+     'True': '#0000ffff',#x1B[39;49;00m#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         ?                     ++#x1B[90m#x1B[39;49;00m#x1B[0m
#x1B[1m#x1B[31mE         #x1B[90m #x1B[39;49;00m }#x1B[90m#x1B[39;49;00m#x1B[0m

#x1B[1m#x1B[.../plotting/legacy/test_plotting_utils.py#x1B[0m:67: AssertionError

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}

[flying-sheep]

Looks like it mostly works except for a small discrepancy in expected values (alpha channel I suppose):

FAILED tests/plotting/legacy/test_plotting_utils.py::test_get_palette_bool_mapping_from_palette_argument - AssertionError: assert {'False': '#f...: '#0000ffff'} == {'False': '#f...e': '#0000ff'}
  
  Differing items:
  {'True': '#0000ffff'} != {'True': '#0000ff'}
  {'False': '#ff0000ff'} != {'False': '#ff0000'}

But this is a lot of added complexity.

Don’t you think it makes more sense to just add both colors in a list if the categorical is boolean?