feat: harden dedupe runtime and NAS deployment
This commit is contained in:
+8
-8
@@ -3,6 +3,7 @@ from __future__ import annotations
|
||||
from pathlib import Path
|
||||
from .acoustic import chromaprint, similarity
|
||||
from .arbitration import choose_winner, is_version_conflict, score
|
||||
from .exact import hardlink_duplicate
|
||||
from .db import connect
|
||||
from .constants import QUARANTINE_SUBDIR, TRASH_SUBDIR
|
||||
|
||||
@@ -22,7 +23,7 @@ def find_candidates(conn,row):
|
||||
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:
|
||||
def acoustic_dedupe(db_path: Path, root: Path, dry_run: bool = False) -> 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()
|
||||
@@ -47,12 +48,11 @@ def acoustic_dedupe(db_path: Path, root: Path) -> dict:
|
||||
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"]))
|
||||
# Confirmed same-version acoustic match keeps the user's logical
|
||||
# path. The physical data is shared via a hard link.
|
||||
if lp.stat().st_ino != wp.stat().st_ino:
|
||||
if not dry_run:
|
||||
hardlink_duplicate(lp, wp)
|
||||
conn.execute("UPDATE audio_tracks SET status='hardlinked',review_status=?,updated_at=CURRENT_TIMESTAMP WHERE id=?", (f"acoustic={sim:.3f};winner={wp}", loser["id"]))
|
||||
stats["equivalent"]+=1
|
||||
return stats
|
||||
|
||||
Reference in New Issue
Block a user