fix: DB sur volume persistant + verrou anti-scrapes parallèles

- DB_PATH → data/arte_dl.db (mappé sur /mnt/user/appdata/arte-dl)
- mkdir data/ au démarrage dans downloader.py et arte_api.py
- asyncio.Lock sur get_all_concerts() : une seule scrape à la fois,
  les requêtes concurrentes attendent le résultat au lieu de relancer

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

arte_api.py

@@ -12,8 +12,20 @@ import tmdb as _tmdb
 logger = logging.getLogger(__name__)
 
 CACHE_TTL = 6 * 3600
-DB_PATH = "arte_dl.db"
+DB_PATH = "data/arte_dl.db"
 _cache: dict = {"data": [], "ts": 0}
+_fetch_lock: asyncio.Lock | None = None
+
+
+def _get_fetch_lock() -> asyncio.Lock:
+    global _fetch_lock
+    if _fetch_lock is None:
+        _fetch_lock = asyncio.Lock()
+    return _fetch_lock
+
+
+import os as _os
+_os.makedirs("data", exist_ok=True)
 
 
 def _db():
@@ -202,7 +214,7 @@ async def get_all_concerts() -> list[dict]:
     if _cache["data"] and now - _cache["ts"] < CACHE_TTL:
         return _cache["data"]
 
-    # Try SQLite cache before hitting the network
+    # Try SQLite cache before acquiring the lock
     db_data, db_ts = _load_db_cache()
     if db_data and now - db_ts < CACHE_TTL:
         logger.info("Concerts cache loaded from DB (%d concerts)", len(db_data))
@@ -210,12 +222,18 @@ async def get_all_concerts() -> list[dict]:
         _cache["ts"] = db_ts
         return _cache["data"]
 
+    # Lock prevents multiple concurrent scrapes (thundering herd)
+    async with _get_fetch_lock():
+        # Re-check after acquiring lock — another coroutine may have populated the cache
+        if _cache["data"] and time.time() - _cache["ts"] < CACHE_TTL:
+            return _cache["data"]
         loop = asyncio.get_event_loop()
         data = await loop.run_in_executor(None, _fetch_all_sync)
         if data:
+            ts = time.time()
             _cache["data"] = data
-        _cache["ts"] = now
-        _save_db_cache(data, now)
+            _cache["ts"] = ts
+            _save_db_cache(data, ts)
     return _cache["data"]
 
 

downloader.py

@@ -11,7 +11,9 @@ import yt_dlp
 from fastapi import BackgroundTasks
 
 OUTPUT_DIR = "/data/Arte"
-DB_PATH = "arte_dl.db"
+DB_PATH = "data/arte_dl.db"
+
+Path("data").mkdir(exist_ok=True)
 
 
 def _db():

tmdb.py

@@ -11,7 +11,7 @@ logger = logging.getLogger(__name__)
 
 _CACHE_DAYS = 30
 _THRESHOLD  = 0.45
-_DB         = "arte_dl.db"
+_DB         = "data/arte_dl.db"
 _IMG_BASE   = "https://image.tmdb.org/t/p"
 _SEARCH_URL = "https://api.themoviedb.org/3/search/movie"