handle None values in ClassifierSettings initialization

src/jabs_postprocess/utils/metadata.py

@@ -26,17 +26,17 @@ class ClassifierSettings:
     def __init__(
         self,
         behavior: str,
-        interpolate: int = DEFAULT_INTERPOLATE,
-        stitch: int = DEFAULT_STITCH,
-        min_bout: int = DEFAULT_MIN_BOUT,
+        interpolate: int | None = None,
+        stitch: int | None = None,
+        min_bout: int | None = None,
     ):
         """Initializes a settings object.
 
         Args:
                 behavior: string containing the name of the behavior
-                interpolate: number of frames where predictions will be interpolated when data is missing
-                stitch: number of frames between "behavior" predictions that will be merged
-                min_bout: minimum number of frames for "behavior" predictions to remain
+                interpolate: number of frames where predictions will be interpolated when data is missing (None uses default)
+                stitch: number of frames between "behavior" predictions that will be merged (None uses default)
+                min_bout: minimum number of frames for "behavior" predictions to remain (None uses default)
 
         Todo:
                 Add back in the functionality to change thresholds (useful for searching for data which has poor/odd probabilities).
@@ -45,9 +45,11 @@ def __init__(
                         threshold_max: high threshold for calling behavior (default 1.0)
         """
         self._behavior = behavior
-        self._interpolate = interpolate
-        self._stitch = stitch
-        self._min_bout = min_bout
+        self._interpolate = (
+            interpolate if interpolate is not None else DEFAULT_INTERPOLATE
+        )
+        self._stitch = stitch if stitch is not None else DEFAULT_STITCH
+        self._min_bout = min_bout if min_bout is not None else DEFAULT_MIN_BOUT
 
     @property
     def behavior(self):

tests/test_generate_behavior_tables.py

@@ -770,3 +770,50 @@ def test_merge_multiple_behavior_tables_empty_group():
 
     with pytest.raises(ValueError, match="No tables provided for behavior: grooming"):
         merge_multiple_behavior_tables(table_groups)
+
+
+# Integration tests without mocks
+
+
+def test_classifier_settings_with_none_values_integration():
+    """Integration test: ClassifierSettings handles None values correctly.
+
+    This is an integration test for issue #45 that verifies the fix works
+    in the actual codebase without mocking. It tests that None values passed
+    to ClassifierSettings are correctly converted to defaults and can be
+    used in comparison operations without raising TypeError.
+
+    This test complements the unit tests in test_metadata.py by verifying
+    the fix works in the context where it was actually failing.
+    """
+    from jabs_postprocess.utils.metadata import (
+        ClassifierSettings,
+        DEFAULT_INTERPOLATE,
+        DEFAULT_STITCH,
+        DEFAULT_MIN_BOUT,
+    )
+    from jabs_postprocess.utils.project_utils import Bouts
+
+    # Create settings with None values (simulates the bug scenario)
+    settings = ClassifierSettings(
+        behavior="grooming",
+        interpolate=None,
+        stitch=None,
+        min_bout=None,
+    )
+
+    # Verify defaults are applied
+    assert settings.interpolate == DEFAULT_INTERPOLATE
+    assert settings.stitch == DEFAULT_STITCH
+    assert settings.min_bout == DEFAULT_MIN_BOUT
+
+    # Create some bout data to test filter_by_settings
+    # This simulates the actual code path where the bug occurred
+    bouts = Bouts.from_value_vector([0, 0, 1, 1, 1, 0, 0, -1, -1, 0, 1, 1])
+
+    # This should not raise TypeError
+    # Previously would fail with: TypeError: '>' not supported between instances of 'NoneType' and 'int'
+    try:
+        bouts.filter_by_settings(settings)
+    except TypeError as e:
+        pytest.fail(f"filter_by_settings raised TypeError with None values: {e}")