🧪 Add test coverage for ChordAnalyzer deduplication and edge cases

services/analysis-engine/tests/test_chords.py

@@ -401,3 +401,61 @@ def test_chord_analyzer_all_n_chords_returns_empty() -> None:
         )
     summary = result["sections"][0]
     assert summary["chords"] == []
+
+
+def test_chord_analyzer_deduplicates_user_chords() -> None:
+    """Test analyzer deduplicates identical user-sourced chords within a section."""
+    analyzer = ChordAnalyzer()
+    sections = [{"id": "verse-1"}]
+    roles_by_section = {
+        "verse-1": [
+            {"harmony": {"chord": "Am", "functionLabel": "vi", "source": "user"}},
+            {"harmony": {"chord": "Am", "functionLabel": "vi repeated", "source": "user"}},
+        ]
+    }
+    result = analyzer.analyze(sections, roles_by_section)
+    assert len(result["sections"][0]["chords"]) == 1
+    assert result["sections"][0]["chords"][0]["chord"] == "Am"
+
+
+def test_chord_analyzer_deduplicates_recognized_chords() -> None:
+    """Test analyzer deduplicates identical recognized chords within a section."""
+    analyzer = ChordAnalyzer()
+    sections = [{"id": "verse-1"}]
+    sr = 22050
+    t = np.linspace(0, 2, sr * 2, endpoint=False)
+    other_stem = np.sin(2 * np.pi * 261.63 * t).astype(np.float32)
+
+    recognized = [
+        {"start_time": 0.0, "end_time": 1.0, "chord": "C", "confidence": "high"},
+        {"start_time": 1.0, "end_time": 2.0, "chord": "C", "confidence": "high"},
+    ]
+
+    with patch.object(analyzer._recognizer, "recognize", return_value=recognized):
+        result = analyzer.analyze(
+            sections,
+            audio_stems={"other": other_stem},
+            sample_rate=sr,
+        )
+
+    assert len(result["sections"][0]["chords"]) == 1
+    assert result["sections"][0]["chords"][0]["chord"] == "C"
+
+
+def test_chord_analyzer_confidence_fallthrough_all_n() -> None:
+    """Test analyzer falls through to medium confidence when recognized chords are all N."""
+    analyzer = ChordAnalyzer()
+
+    # Setup inputs for _compute_section_confidence directly
+    chords = [{"chord": "C", "functionLabel": "", "source": "model"}]
+    recognized_chords = [{"start_time": 0.0, "end_time": 1.0, "chord": "N", "confidence": "high"}]
+    user_chords = []
+
+    confidence_level, confidence_source = analyzer._compute_section_confidence(
+        chords=chords,
+        recognized_chords=recognized_chords,
+        user_chords=user_chords,
+    )
+
+    assert confidence_level == "medium"
+    assert confidence_source == "model"