Cleanup & add tests for CSV loading

.gitignore

@@ -82,3 +82,7 @@ venv/
 
 # written by setuptools_scm
 **/_version.py
+
+# Files produced during testing
+*.csv
+*.clf

example/feature_visualization_example.py

@@ -16,10 +16,12 @@
 d = {
     "test": [-100, 200, 300, 500, 900, 300],
     "label": [1, 2, 3, 4, 5, 6],
+    "index": [1, 2, 3, 4, 5, 6],
     "feature1": [100, 200, 300, 500, 900, 1001],
     "feature2": [2200, 2100, 2000, 1500, 1300, 1001],
 }
 df = pd.DataFrame(data=d)
+# df.to_csv("example_data.csv", index=False)
 
 viewer = napari.Viewer()
 label_layer = viewer.add_labels(lbl_img_np)

pyproject.toml

@@ -31,6 +31,7 @@ dependencies = [
     "scikit-image",
     "matplotlib",
     "pandas",
+    "packaging",
 ]
 
 [project.optional-dependencies]

src/napari_feature_visualization/_tests/test_feature_vis.py

@@ -1,7 +1,8 @@
-import importlib
-
+import napari
 import numpy as np
 import pandas as pd
+import pytest
+from packaging import version
 
 from napari_feature_visualization.feature_vis import feature_vis
 
@@ -18,55 +19,93 @@ def create_label_img():
     return lbl_img_np
 
 
-def test_feature_vis_widget(make_napari_viewer):
-    lbl_img = create_label_img()
-
-    # Dummy df for this test
+def create_feature_df():
     d = {
         "test": [-100, 200, 300, 500, 900, 300],
         "label": [1, 2, 3, 4, 5, 6],
+        "index": [1, 2, 3, 4, 5, 6],
         "feature1": [100, 200, 300, 500, 900, 1001],
         "feature2": [2200, 2100, 2000, 1500, 1300, 1001],
     }
     df = pd.DataFrame(data=d)
+    # Ensure index is the labels for correct matching
+    return df
+
 
+@pytest.mark.parametrize(
+    "load_features_from", ["CSV File", "Layer Properties"]
+)
+def test_feature_vis_widget(make_napari_viewer, load_features_from):
+    lbl_img = create_label_img()
+    df = create_feature_df()
     viewer = make_napari_viewer()
     label_layer = viewer.add_labels(lbl_img)
-    label_layer.features = df
 
     feature_vis_widget = feature_vis()
 
-    # FIXME: It appears feature test is used. How do I test other feature
-    # selection? Setting it in the function appears to have no effect.
-
-    # if we "call" this object, it'll execute our function
-    feature_vis_widget(
-        label_layer=label_layer,
-        load_features_from="Layer Properties",
-        feature="feature1",
-    )
-
-    # Test differently depending on napari version, as colormap class has
-    # changed
-    print(
-        importlib.util.find_spec("napari.utils.colormaps.DirectLabelColormap")
-    )
-    colormaps_module = importlib.import_module("napari.utils.colormaps")
-    DirectLabelColormap = getattr(
-        colormaps_module, "DirectLabelColormap", None
-    )
-    if DirectLabelColormap is not None:
-        # napari >= 0.4.19 tests
-        from napari.utils.colormaps import DirectLabelColormap
+    if load_features_from == "CSV File":
+        df.to_csv("example_data.csv", index=False)
+        # if we "call" this object, it'll execute our function
+        feature_vis_widget(
+            label_layer=label_layer,
+            load_features_from=load_features_from,
+            DataFrame="example_data.csv",
+            label_column="label",
+            feature="feature1",
+        )
+    elif load_features_from == "Layer Properties":
+        label_layer.features = df
+        # if we "call" this object, it'll execute our function
+        feature_vis_widget(
+            label_layer=label_layer,
+            load_features_from=load_features_from,
+            feature="feature1",
+            label_column="label",
+        )
 
+    napari_version = version.parse(napari.__version__)
+    if napari_version >= version.parse("0.4.19"):
+        assert len(label_layer.colormap.color_dict) == 8
         np.testing.assert_array_almost_equal(
             label_layer.colormap.color_dict[3],
             np.array([0.229739, 0.322361, 0.545706, 1.0]),
         )
     else:
-        # napari < 0.4.19 test
-        assert len(label_layer.colormap.colors) == 6
+        assert len(label_layer.colormap.colors) == 7
         np.testing.assert_array_almost_equal(
-            label_layer.colormap.colors[2],
+            label_layer.colormap.colors[3],
             np.array([0.229739, 0.322361, 0.545706, 1.0]),
         )
+
+
+# def test_feature_vis_from_csv(make_napari_viewer):
+#     lbl_img = create_label_img()
+#     df = create_feature_df()
+
+#     viewer = make_napari_viewer()
+#     label_layer = viewer.add_labels(lbl_img)
+#     label_layer.features = df
+
+#     feature_vis_widget = feature_vis()
+
+#     # if we "call" this object, it'll execute our function
+#     feature_vis_widget(
+#         label_layer=label_layer,
+#         load_features_from="CSV File",
+#         DataFrame="example_data.csv",
+#         feature="feature1",
+#     )
+
+#     napari_version = version.parse(napari.__version__)
+#     if napari_version >= version.parse("0.4.19"):
+#         assert len(label_layer.colormap.color_dict) == 8
+#         np.testing.assert_array_almost_equal(
+#             label_layer.colormap.color_dict[3],
+#             np.array([0.229739, 0.322361, 0.545706, 1.0]),
+#         )
+#     else:
+#         assert len(label_layer.colormap.colors) == 7
+#         np.testing.assert_array_almost_equal(
+#             label_layer.colormap.colors[3],
+#             np.array([0.229739, 0.322361, 0.545706, 1.0]),
+#         )

src/napari_feature_visualization/feature_vis.py

@@ -2,15 +2,26 @@
 
 import pathlib
 
-import matplotlib.pyplot as plt
+import matplotlib
 import napari
 import numpy as np
 import pandas as pd
 from magicgui import magic_factory
+from packaging import version
 
 from napari_feature_visualization.utils import ColormapChoices, get_df
 
 
+def check_default_label_column(df):
+    if "label" in df:
+        return "label"
+    elif "Label" in df:
+        return "Label"
+    elif "index" in df:
+        return "index"
+    return ""
+
+
 def _init(widget):
     def get_feature_choices(*args):
         if widget.load_features_from.value == "CSV File":
@@ -41,12 +52,7 @@ def update_df_columns(event):
         widget.feature.reset_choices()
         widget.label_column.reset_choices()
         features = widget.feature.choices
-        if "label" in features:
-            widget.label_column.value = "label"
-        elif "Label" in features:
-            widget.label_column.value = "Label"
-        elif "index" in features:
-            widget.label_column.value = "index"
+        widget.label_column.value = check_default_label_column(features)
 
         # if load_features_from is toggled, make the widget.DataFrame disappear
         if widget.load_features_from.value == "Layer Properties":
@@ -109,15 +115,18 @@ def feature_vis(
         site_df = get_df(DataFrame)
     else:
         site_df = pd.DataFrame(label_layer.properties)
-        label_column = "label"
+
+    if label_column == "":
+        label_column = check_default_label_column(site_df)
 
     site_df.loc[:, "label"] = site_df[str(label_column)].astype(int)
     # Check that there is one unique label for every entry in the dataframe
-    # => It's a site dataframe, not one containing many different sites
-    # TODO: How to feedback this issue to the user?
-    assert len(site_df["label"].unique()) == len(
-        site_df
-    ), "A feature dataframe with non-unique labels was provided. The visualize_feature_on_label_layer function is not designed for this."
+    if len(site_df["label"].unique()) != len(site_df):
+        raise ValueError(
+            "A feature dataframe with non-unique labels was provided. The "
+            "visualize_feature_on_label_layer function is not designed for "
+            "this."
+        )
     # Rescale feature between 0 & 1 to make a colormap
     site_df["feature_scaled"] = (site_df[feature] - lower_contrast_limit) / (
         upper_contrast_limit - lower_contrast_limit
@@ -126,7 +135,9 @@ def feature_vis(
     site_df.loc[site_df["feature_scaled"] < 0, "feature_scaled"] = 0
     site_df.loc[site_df["feature_scaled"] > 1, "feature_scaled"] = 1
 
-    colors = plt.cm.get_cmap(Colormap.value)(site_df["feature_scaled"])
+    colors = matplotlib.colormaps.get_cmap(Colormap.value)(
+        site_df["feature_scaled"]
+    )
 
     # Create an array where the index is the label value and the value is
     # the feature value
@@ -135,13 +146,16 @@ def feature_vis(
     label_properties = {feature: np.round(properties_array, decimals=2)}
 
     colormap = dict(zip(site_df["label"], colors))
-    colormap[None] = [0.0, 0.0, 0.0, 0.0]
-    # If in napari >= 0.4.19, use DirectLabelColormap
-    try:
+    # Show missing objects as black
+    colormap[None] = [0.0, 0.0, 0.0, 1.0]
+
+    # Handle different colormap APIs depending on the napari version
+    napari_version = version.parse(napari.__version__)
+    if napari_version >= version.parse("0.4.19"):
         from napari.utils.colormaps import DirectLabelColormap
 
         label_layer.colormap = DirectLabelColormap(color_dict=colormap)
-    except ImportError:
+    else:
         label_layer.color = colormap
 
     if load_features_from == "CSV File":