[bacescuandrei]

Gradient.compute_color() looks scores up in a precomputed curve. The curve's length depends on how evenly maxValue - minValue divides between the color stops. This causes three problems:

• With four or more colors and a range such as 0–11, scores near the maximum raise IndexError. This aborts ToExcel.to_xlsx() and ToSvg.to_svg() for the whole layer.
• Each curve segment repeats its end color, so intermediate scores pick up the next stop early. With red → yellow → green over 0–4, a score of 1 renders as yellow.
• A single-color gradient raises ZeroDivisionError.

This change interpolates linearly in RGB between evenly spaced stops, approximately as the ATT&CK Navigator does. Navigator uses tinygradient's RGB interpolation, quantised to 100 steps, so small differences remain on steep gradients. It also removes the precomputed curve attribute and _compute_curve. Neither is documented or used elsewhere in the package.

Validation:

• New tests/test_gradient.py with 13 cases: 9 of them fail on main and all pass with the change.
• uv run pytest: 310 passed, 4 skipped.
• ruff check and ruff format --check: clean.
• Reproduced the export crash end to end with a local ATT&CK v20.0 bundle: a layer with gradient #ff0000, #00ff00, #0000ff, #ffffff over 0–11 and T1003 scored 10. On main, both to_xlsx and to_svg raise IndexError; with the change, both succeed.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud