feat: add confidence scoring

2026-07-10 18:26:52 +02:00
parent 1c013dfdd1
commit 3888d2ab8a
2 changed files with 58 additions and 0 deletions
@@ -0,0 +1,37 @@
+from lyra_worker.confidence import score_confidence
+from lyra_worker.types import Candidate, MBTarget, Quality
+
+Q = Quality(fmt="FLAC", lossless=True, bit_depth=16, sample_rate=44100)
+
+
+def _cand(artist, album, tracks=10):
+    return Candidate(source="qobuz", source_ref="x", matched_artist=artist,
+                     matched_album=album, quality=Q, track_count=tracks)
+
+
+def test_exact_match_scores_high():
+    target = MBTarget(artist="Radiohead", album="In Rainbows", track_count=10)
+    assert score_confidence(target, _cand("Radiohead", "In Rainbows", 10)) > 0.95
+
+
+def test_case_and_whitespace_insensitive():
+    target = MBTarget(artist="Radiohead", album="In Rainbows")
+    assert score_confidence(target, _cand("  radiohead ", "IN RAINBOWS")) > 0.95
+
+
+def test_wrong_album_scores_low():
+    target = MBTarget(artist="Radiohead", album="In Rainbows", track_count=10)
+    assert score_confidence(target, _cand("Radiohead", "OK Computer", 12)) < 0.7
+
+
+def test_unknown_track_count_does_not_penalize():
+    target = MBTarget(artist="Radiohead", album="In Rainbows", track_count=None)
+    # track_count unknown -> track_factor neutral; name match still drives a high score
+    assert score_confidence(target, _cand("Radiohead", "In Rainbows", 99)) > 0.95
+
+
+def test_track_count_mismatch_lowers_score():
+    target = MBTarget(artist="Radiohead", album="In Rainbows", track_count=10)
+    good = score_confidence(target, _cand("Radiohead", "In Rainbows", 10))
+    bad = score_confidence(target, _cand("Radiohead", "In Rainbows", 4))
+    assert bad < good