Add explainable matching engine

serato_doctor/matching.py

@@ -0,0 +1,86 @@
+import re
+import unicodedata
+from collections import defaultdict
+from pathlib import Path
+from typing import DefaultDict, Iterable, List, Tuple
+
+from serato_doctor.models.match import MatchEvidence, TrackMatch
+from serato_doctor.models.reference import TrackReference
+from serato_doctor.models.track import DiskTrack
+
+
+def normalize(value: str) -> str:
+    return unicodedata.normalize("NFKC", value).casefold()
+
+
+def cloud_conflict_name(filename: str) -> str:
+    """Remove a trailing numeric cloud-conflict suffix from a filename stem."""
+
+    path = Path(filename)
+    stem = re.sub(r" \d+$", "", path.stem)
+    return normalize(stem + path.suffix)
+
+
+def score_candidate(reference: TrackReference, track: DiskTrack) -> TrackMatch:
+    reference_name = reference.filename
+    track_name = track.filename
+
+    if reference_name == track_name:
+        filename_points = 60
+        filename_reason = "Filename is identical"
+    elif normalize(reference_name) == normalize(track_name):
+        filename_points = 55
+        filename_reason = "Filename matches after case and Unicode normalization"
+    elif cloud_conflict_name(reference_name) == cloud_conflict_name(track_name):
+        filename_points = 50
+        filename_reason = "Filename matches after removing a numeric conflict suffix"
+    else:
+        filename_points = 0
+        filename_reason = "Filename does not match"
+
+    same_extension = normalize(reference.path.suffix) == normalize(track.suffix)
+    same_parent = normalize(reference.path.parent.name) == normalize(
+        track.path.parent.name
+    )
+    evidence = (
+        MatchEvidence(
+            "filename",
+            filename_points > 0,
+            filename_points,
+            60,
+            filename_reason,
+        ),
+        MatchEvidence(
+            "extension",
+            same_extension,
+            10 if same_extension else 0,
+            10,
+            "File extension matches" if same_extension else "File extension differs",
+        ),
+        MatchEvidence(
+            "parent_folder",
+            same_parent,
+            20 if same_parent else 0,
+            20,
+            "Parent folder matches" if same_parent else "Parent folder differs",
+        ),
+    )
+    return TrackMatch(reference, track, evidence)
+
+
+class MatchingEngine:
+    """Find and rank filename-related disk candidates without modifying files."""
+
+    def __init__(self, tracks: Iterable[DiskTrack]):
+        self._by_conflict_name: DefaultDict[str, List[DiskTrack]] = defaultdict(list)
+        for track in tracks:
+            self._by_conflict_name[cloud_conflict_name(track.filename)].append(track)
+
+    def candidates_for(self, reference: TrackReference) -> Tuple[TrackMatch, ...]:
+        candidates = self._by_conflict_name.get(
+            cloud_conflict_name(reference.filename), []
+        )
+        matches = [score_candidate(reference, track) for track in candidates]
+        return tuple(
+            sorted(matches, key=lambda match: (-match.score, str(match.track.path)))
+        )