geotiff: backend-parity tests for bool nodata rejection (#1921)

[brendancol]

Summary

• Add backend-parity tests for the to_geotiff bool-nodata guard from geotiff: to_geotiff(nodata=True) silently writes unparseable "True" tag #1911. write_vrt and direct write_geotiff_gpu calls were missed by the original fix.
• Four xfail(strict=True) tests on write_vrt capture the fix surface for issue geotiff: write_vrt(nodata=True) silently writes 'True' string into VRT XML #1921. One passing test pins the current buggy <NoDataValue>True</NoDataValue> emission so the source fix shows up as a diff.
• write_geotiff_gpu direct-call tests pin the existing rejection (driven by the build_geo_tags defense-in-depth guard) so a future refactor that moves that check surfaces here instead of in user code.

Test-only PR; source fix tracked in #1921. Found by /sweep-test-coverage pass 15.

Test plan

• pytest xrspatial/geotiff/tests/test_write_vrt_bool_nodata_1921.py -> 13 passed, 4 xfailed locally
• pytest xrspatial/geotiff/tests/test_nodata_bool_rejection_1911.py still green (no overlap regression)

[Copilot]

Pull request overview

Adds test coverage to ensure GeoTIFF/VRT writer entry points consistently reject nodata=True/False (and np.bool_) by raising TypeError, and records the sweep-test-coverage state update for issue #1921 tracking.

Changes:

• Add new backend-parity tests covering write_vrt bool-nodata behavior (including strict xfails for the expected future fix).
• Add GPU-path tests pinning bool-nodata rejection for direct write_geotiff_gpu calls and the to_geotiff(gpu=True) dispatch path.
• Update .claude/sweep-test-coverage-state.csv to record sweep pass 15 / issue #1921 metadata.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
xrspatial/geotiff/tests/test_write_vrt_bool_nodata_1921.py	Introduces new parity tests for bool/np.bool_ nodata rejection across write_vrt and GPU writer paths.
.claude/sweep-test-coverage-state.csv	Updates sweep-test-coverage tracking state for geotiff module / issue #1921.

Comments suppressed due to low confidence (1)

xrspatial/geotiff/tests/test_write_vrt_bool_nodata_1921.py:145

• These tests should be skipped with the repo-standard GPU marker (reason "cupy + CUDA required") rather than only checking _HAS_CUPY with reason "cupy not available". As written, they may run (and fail) when CuPy is installed but CUDA is unavailable; consider replacing with @requires_gpu from geotiff/tests/conftest.py.

@pytest.mark.skipif(not _HAS_CUPY, reason="cupy not available")
@pytest.mark.parametrize(
    "bad",
    [True, False, np.bool_(True), np.bool_(False)],
)
def test_write_geotiff_gpu_rejects_bool_nodata(uint8_da, tmp_path, bad):
    """Direct ``write_geotiff_gpu`` call rejects bool nodata.

    Currently raises ``TypeError`` only because the deeper
    ``build_geo_tags`` guard fires. Pinning the behaviour so a refactor
    that drops the deeper guard surfaces here, not in user code.
    """
    from xrspatial.geotiff import write_geotiff_gpu
    path = str(tmp_path / "gpu_1921_bad.tif")
    with pytest.raises(TypeError, match="nodata must be numeric"):
        write_geotiff_gpu(uint8_da, path, nodata=bad)


@pytest.mark.skipif(not _HAS_CUPY, reason="cupy not available")
def test_to_geotiff_gpu_dispatch_rejects_bool_nodata(uint8_da, tmp_path):

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.