feat(classifier): add _cosine() helper for embedding similarity

scripts/classifier_adapters.py

@@ -7,6 +7,7 @@ from __future__ import annotations
 
 import abc
 from collections import defaultdict
+from pathlib import Path
 from typing import Any
 
 __all__ = [
@@ -117,6 +118,14 @@ def compute_metrics(
     return result
 
 
+
+def _cosine(a: list[float], b: list[float]) -> float:
+    dot = sum(x * y for x, y in zip(a, b))
+    norm_a = sum(x * x for x in a) ** 0.5
+    norm_b = sum(x * x for x in b) ** 0.5
+    return dot / (norm_a * norm_b) if norm_a and norm_b else 0.0
+
+
 class ClassifierAdapter(abc.ABC):
     """Abstract base for all email classifier adapters."""
 

tests/test_classifier_adapters.py

@@ -268,3 +268,28 @@ def test_finetuned_adapter_unload_clears_pipeline():
         assert adapter._pipeline is not None
         adapter.unload()
         assert adapter._pipeline is None
+
+# ---- _cosine() tests ----
+
+def test_cosine_identical_unit_vectors():
+    import math
+    from scripts.classifier_adapters import _cosine
+    assert _cosine([1.0, 0.0], [1.0, 0.0]) == pytest.approx(1.0)
+
+
+def test_cosine_orthogonal_vectors():
+    from scripts.classifier_adapters import _cosine
+    assert _cosine([1.0, 0.0], [0.0, 1.0]) == pytest.approx(0.0)
+
+
+def test_cosine_known_value():
+    import math
+    from scripts.classifier_adapters import _cosine
+    # [1,0] vs [1/sqrt(2), 1/sqrt(2)] → dot = 1/sqrt(2), both norms = 1 → 1/sqrt(2)
+    v = [1.0 / math.sqrt(2), 1.0 / math.sqrt(2)]
+    assert _cosine([1.0, 0.0], v) == pytest.approx(1.0 / math.sqrt(2))
+
+
+def test_cosine_zero_vector_returns_zero():
+    from scripts.classifier_adapters import _cosine
+    assert _cosine([0.0, 0.0], [1.0, 0.0]) == pytest.approx(0.0)