feat: add acoustic dedupe pipeline and docker runtime

am_deduper/l2.py

@@ -0,0 +1,58 @@
+from __future__ import annotations
+
+from pathlib import Path
+from .acoustic import chromaprint, similarity
+from .arbitration import choose_winner, is_version_conflict, score
+from .db import connect
+from .constants import QUARANTINE_SUBDIR, TRASH_SUBDIR
+
+MIN_SIMILARITY = 0.90
+HIGH_SIMILARITY = 0.95
+REVIEW_SIMILARITY = 0.80
+DURATION_DELTA = 3.0
+
+
+def _set_meta(conn, row, fp):
+    conn.execute("UPDATE audio_tracks SET chromaprint=?,score=?,updated_at=CURRENT_TIMESTAMP WHERE id=?", (fp,score(row),row["id"]))
+
+
+def find_candidates(conn,row):
+    d=row["duration"]
+    if d is None: return []
+    return conn.execute("SELECT * FROM audio_tracks WHERE id!=? AND status IN ('active','hardlinked') AND duration IS NOT NULL AND duration BETWEEN ? AND ?", (row["id"],d-DURATION_DELTA,d+DURATION_DELTA)).fetchall()
+
+
+def acoustic_dedupe(db_path: Path, root: Path) -> dict:
+    stats={"checked":0,"equivalent":0,"review":0,"unique":0,"errors":0}
+    with connect(db_path) as conn:
+        rows=conn.execute("SELECT * FROM audio_tracks WHERE status='active' ORDER BY id").fetchall()
+        for row in rows:
+            if row["chromaprint"]: continue
+            fp=chromaprint(Path(row["file_path"])); stats["checked"]+=1
+            if not fp: stats["errors"]+=1; continue
+            _set_meta(conn,row,fp)
+            candidates=find_candidates(conn,row); best=None
+            for c in candidates:
+                if not c["chromaprint"]:
+                    cfp=chromaprint(Path(c["file_path"]))
+                    if cfp: _set_meta(conn,c,cfp); c=dict(c); c["chromaprint"]=cfp
+                sim=similarity(fp,c["chromaprint"] or "")
+                if sim>=REVIEW_SIMILARITY and (best is None or sim>best[0]): best=(sim,c)
+            if not best or best[0]<REVIEW_SIMILARITY:
+                stats["unique"]+=1; continue
+            sim,c=best
+            if sim<MIN_SIMILARITY or is_version_conflict(row,c):
+                conn.execute("UPDATE audio_tracks SET status='review',review_status=?,updated_at=CURRENT_TIMESTAMP WHERE id=?",(f"similarity={sim:.3f}",row["id"]))
+                stats["review"]+=1; continue
+            winner=choose_winner([row,c]); loser=c if winner["id"]==row["id"] else row
+            lp=Path(loser["file_path"]); wp=Path(winner["file_path"])
+            if lp.exists() and wp.exists():
+                trash=root/TRASH_SUBDIR
+                trash.mkdir(parents=True,exist_ok=True)
+                target=trash/lp.name
+                n=1
+                while target.exists(): target=trash/f"{lp.stem}.{n}{lp.suffix}"; n+=1
+                lp.replace(target)
+                conn.execute("UPDATE audio_tracks SET status='trashed',file_path=?,trashed_at=CURRENT_TIMESTAMP,review_status=?,updated_at=CURRENT_TIMESTAMP WHERE id=?",(str(target),f"equivalent={sim:.3f};winner={wp}",loser["id"]))
+                stats["equivalent"]+=1
+    return stats