Merge pull request #56 from SystemsGenetics/starch-tests

Add comprehensive unit tests for StarchArea

Author: AdenAthar

tests/test_Analyses/test_StarchArea.py

@@ -1,15 +1,221 @@
-from Granny.Analyses.StarchArea import StarchArea
-from Granny.Models.Images.RGBImage import RGBImage
-from Granny.Models.IO.RGBImageFile import RGBImageFile
-from Granny.Models.Values.StringValue import StringValue
-from Granny.Models.Values.ImageListValue import ImageListValue
+import numpy as np
+import pytest
+from Granny.Analyses.StarchArea import StarchArea, StarchScales, load_starch_scales
 
 
-def test_StarchAnalyses():
+# ---------------------------------------------------------------------------
+# load_starch_scales / StarchScales
+# ---------------------------------------------------------------------------
+
+def test_load_starch_scales_returns_dict():
+    """YAML loads as a non-empty dictionary."""
+    data = load_starch_scales()
+    assert isinstance(data, dict)
+    assert len(data) > 0
+
+
+def test_load_starch_scales_expected_varieties():
+    """All expected varieties are present in the YAML."""
+    data = load_starch_scales()
+    expected = {
+        "HONEYCRISP", "WA38_1", "WA38_2", "ENZA",
+        "CORNELL", "PURDUE", "DANJOU",
+        "GOLDEN_DELICIOUS", "GRANNY_SMITH", "JONAGOLD",
+    }
+    assert expected.issubset(data.keys())
+
+
+def test_load_starch_scales_index_rating_same_length():
+    """Every variety has equal-length index and rating lists."""
+    data = load_starch_scales()
+    for variety, scale in data.items():
+        assert len(scale["index"]) == len(scale["rating"]), (
+            f"{variety}: index length {len(scale['index'])} != "
+            f"rating length {len(scale['rating'])}"
+        )
+
+
+def test_load_starch_scales_ratings_between_0_and_1():
+    """All rating values are valid percentages (0.0-1.0)."""
+    data = load_starch_scales()
+    for variety, scale in data.items():
+        for r in scale["rating"]:
+            assert 0.0 <= r <= 1.0, f"{variety} has out-of-range rating: {r}"
+
+
+def test_starch_scales_class_attributes():
+    """StarchScales class exposes variety names as attributes."""
+    data = load_starch_scales()
+    for variety in data:
+        assert hasattr(StarchScales, variety), f"StarchScales missing attribute: {variety}"
+
+
+# ---------------------------------------------------------------------------
+# StarchArea initialisation & parameters
+# ---------------------------------------------------------------------------
+
+def test_starch_area_instantiation():
+    analysis = StarchArea()
+    assert analysis is not None
+
+
+def test_default_threshold_is_140():
+    analysis = StarchArea()
+    assert analysis.starch_threshold.getValue() == 140
+
+
+def test_threshold_accepts_boundary_values():
+    """Threshold should accept the full 0-255 range without error."""
     analysis = StarchArea()
-    # Set up input images from demo directory
-    analysis.input_images.setValue("demo/cross_section_images/full_masked_images")
+    analysis.starch_threshold.setValue(0)
+    assert analysis.starch_threshold.getValue() == 0
+    analysis.starch_threshold.setValue(255)
+    assert analysis.starch_threshold.getValue() == 255
+
+
+def test_default_blur_kernel_is_7():
+    analysis = StarchArea()
+    assert analysis.blur_kernel.getValue() == 7
+
+
+def test_default_mask_alpha_is_0_6():
+    analysis = StarchArea()
+    assert analysis.mask_alpha.getValue() == pytest.approx(0.6)
+
+
+# ---------------------------------------------------------------------------
+# _calculateStarch - synthetic image tests
+# ---------------------------------------------------------------------------
+
+def _make_bgr(gray_value: int, size: int = 100) -> np.ndarray:
+    """Create a solid-colour BGR image with a given grayscale brightness."""
+    channel = np.full((size, size), gray_value, dtype=np.uint8)
+    return np.stack([channel, channel, channel], axis=-1)
+
+
+def test_calculate_starch_all_dark_returns_high_ratio():
+    """A very dark image should be almost entirely starch."""
+    analysis = StarchArea()
+    img = _make_bgr(10)
+    ratio, _ = analysis._calculateStarch(img)
+    assert ratio > 0.8, f"Expected ratio > 0.8 for dark image, got {ratio}"
+
+
+def test_calculate_starch_bright_dominant_image_returns_low_ratio():
+    """An image that is mostly bright should detect low starch."""
+    size = 100
+    img = np.full((size, size, 3), 240, dtype=np.uint8)
+    # Small dark patch (10x10) in the corner - starch
+    img[:10, :10] = 10
+
+    analysis = StarchArea()
+    ratio, _ = analysis._calculateStarch(img)
+    assert ratio < 0.2, f"Expected ratio < 0.2 for mostly-bright image, got {ratio}"
+
 
-    # This test just verifies that StarchArea can be instantiated
-    # and that input_images can be set without errors
-    assert analysis.input_images.getValue() == "demo/cross_section_images/full_masked_images"
+def test_calculate_starch_returns_ratio_between_0_and_1():
+    """Starch ratio must always be a valid proportion."""
+    analysis = StarchArea()
+    img = _make_bgr(128)
+    ratio, _ = analysis._calculateStarch(img)
+    assert 0.0 <= ratio <= 1.0
+
+
+def test_calculate_starch_returns_image_same_shape():
+    """The returned image must have the same shape as the input."""
+    analysis = StarchArea()
+    img = _make_bgr(100, size=64)
+    _, result = analysis._calculateStarch(img)
+    assert result.shape == img.shape
+
+
+def test_calculate_starch_higher_threshold_gives_higher_ratio():
+    """Raising the threshold should classify more pixels as starch."""
+    img = _make_bgr(150)
+    low_analysis = StarchArea()
+    low_analysis.starch_threshold.setValue(80)   # ~31%
+    high_analysis = StarchArea()
+    high_analysis.starch_threshold.setValue(200)  # ~78%
+
+    low_ratio, _ = low_analysis._calculateStarch(img)
+    high_ratio, _ = high_analysis._calculateStarch(img)
+    assert high_ratio >= low_ratio
+
+
+def test_calculate_starch_mixed_image():
+    """An image split between dark and bright halves should give ~50% ratio."""
+    size = 100
+    img = np.zeros((size, size, 3), dtype=np.uint8)
+    img[:, :50] = 10    # left half: very dark (starch)
+    img[:, 50:] = 240   # right half: very bright (no starch)
+
+    analysis = StarchArea()
+    ratio, _ = analysis._calculateStarch(img)
+    assert 0.3 < ratio < 0.7, f"Expected roughly 0.5, got {ratio}"
+
+
+# ---------------------------------------------------------------------------
+# _calculateIndex
+# ---------------------------------------------------------------------------
+
+def test_calculate_index_returns_all_varieties():
+    """_calculateIndex should return a key for every loaded variety."""
+    analysis = StarchArea()
+    data = load_starch_scales()
+    results = analysis._calculateIndex(0.5)
+    assert set(results.keys()) == set(data.keys())
+
+
+def test_calculate_index_exact_match():
+    """When target exactly matches a rating, that index is selected."""
+    analysis = StarchArea()
+    # CORNELL index 5 -> rating 0.537191447
+    target = 0.537191447
+    results = analysis._calculateIndex(target)
+    assert results["CORNELL"] == pytest.approx(5.0)
+
+
+def test_calculate_index_closest_match():
+    """A target slightly off a known rating should still pick the nearest index."""
+    analysis = StarchArea()
+    # HONEYCRISP index 1 -> rating 0.981640465
+    results = analysis._calculateIndex(0.982)
+    assert results["HONEYCRISP"] == pytest.approx(1.0)
+
+
+def test_calculate_index_values_are_floats():
+    """All returned index values should be numeric."""
+    analysis = StarchArea()
+    results = analysis._calculateIndex(0.5)
+    for variety, index in results.items():
+        assert isinstance(index, (int, float)), f"{variety} index is not numeric"
+
+
+# ---------------------------------------------------------------------------
+# _drawMask
+# ---------------------------------------------------------------------------
+
+def test_draw_mask_zeros_darken_pixels():
+    """Pixels where mask == 0 should be darkened."""
+    analysis = StarchArea()
+    img = _make_bgr(200, size=10)
+    mask = np.zeros((10, 10), dtype=np.uint8)  # all starch
+    result = analysis._drawMask(img, mask)
+    assert result.mean() < img.mean()
+
+
+def test_draw_mask_ones_leave_pixels_unchanged():
+    """Pixels where mask == 1 should be unchanged."""
+    analysis = StarchArea()
+    img = _make_bgr(200, size=10)
+    mask = np.ones((10, 10), dtype=np.uint8)  # no starch
+    result = analysis._drawMask(img, mask)
+    np.testing.assert_array_equal(result, img)
+
+
+def test_draw_mask_output_shape_matches_input():
+    analysis = StarchArea()
+    img = _make_bgr(128, size=50)
+    mask = np.ones((50, 50), dtype=np.uint8)
+    result = analysis._drawMask(img, mask)
+    assert result.shape == img.shape