Update src/utils.py

src/utils.py

@@ -14,11 +14,78 @@ ulog = logging.getLogger("utils")
 
 # Repo root (/home/user/app)
 APP_DIR = Path(__file__).resolve().parents[1]
+PERSIST = Path("/data/olist.sqlite")                   # preferred persistent path
+FALLBACK = APP_DIR / "olist.sqlite"                    # non-persistent fallback
+ENV_DB = os.environ.get("DB_PATH")  
 
 # Single source of truth for DB location:
 # - Default to Hugging Face persistent volume at /data
 # - Can be overridden via env var DB_PATH if needed
-DB_PATH = Path(os.environ.get("DB_PATH", "/data/olist.sqlite"))
+#DB_PATH = Path(os.environ.get("DB_PATH", "/data/olist.sqlite"))
+
+def _can_write_dir(d: Path) -> bool:
+    try:
+        d.mkdir(parents=True, exist_ok=True)
+        test = d / ".perm_test"
+        with open(test, "wb") as f:
+            f.write(b"ok")
+        test.unlink()
+        return True
+    except Exception as e:
+        ulog.warning("perm: cannot write to %s: %s", d, e)
+        return False
+
+def _choose_path() -> Path:
+    if ENV_DB:
+        p = Path(ENV_DB).resolve()
+        # if file exists OR parent is writable, we’ll use it
+        if p.exists() or _can_write_dir(p.parent):
+            return p
+        ulog.warning("db: DB_PATH=%s not writable; falling back…", p)
+    # prefer /data if writable or already exists
+    if PERSIST.exists() or _can_write_dir(PERSIST.parent):
+        return PERSIST
+    # last resort: fallback in app dir
+    ulog.warning("db: /data not writable; using non-persistent %s", FALLBACK)
+    return FALLBACK
+
+def _seed_if_missing(path: Path):
+    if path.exists():
+        return
+    seed = APP_DIR / "olist.sqlite"    # optional repo copy
+    try:
+        path.parent.mkdir(parents=True, exist_ok=True)
+    except Exception as e:
+        ulog.exception("db: mkdir failed for %s: %s", path.parent, e)
+    if seed.exists():
+        try:
+            shutil.copy2(seed, path)
+            ulog.info("db: seeded %s from %s", path, seed)
+        except Exception as e:
+            ulog.exception("db: seed copy failed %s -> %s: %s", seed, path, e)
+    else:
+        ulog.warning("db: no seed found at %s; creating empty DB at %s", seed, path)
+
+def get_connection():
+    path = _choose_path()
+    _seed_if_missing(path)
+    try:
+        conn = sqlite3.connect(path, check_same_thread=False, timeout=30)
+        conn.execute("PRAGMA foreign_keys = ON;")
+        try:
+            size = path.stat().st_size if path.exists() else 0
+        except Exception:
+            size = -1
+        ulog.debug("db: open path=%s exists=%s size=%s bytes", path, path.exists(), size)
+        try:
+            dblist = conn.execute("PRAGMA database_list").fetchall()
+            ulog.debug("db: PRAGMA database_list => %s", dblist)
+        except Exception as e:
+            ulog.debug("db: PRAGMA database_list failed: %s", e)
+        return conn
+    except Exception as e:
+        ulog.exception("db: connect failed for %s: %s", path, e)
+        raise
 
 def _ensure_db_file():
     """
@@ -44,28 +111,3 @@ def _ensure_db_file():
     else:
         ulog.warning("db: no seed DB found at %s; will create empty DB at %s", seed, DB_PATH)
 
-def get_connection():
-    """
-    Open a connection to the persistent DB.
-    Logs path and PRAGMA database_list so you can verify which file is used.
-    """
-    _ensure_db_file()
-    try:
-        conn = sqlite3.connect(DB_PATH, check_same_thread=False, timeout=30)
-        conn.execute("PRAGMA foreign_keys = ON;")
-        # Diagnostics
-        try:
-            size = DB_PATH.stat().st_size if DB_PATH.exists() else 0
-        except Exception:
-            size = -1
-        ulog.debug("db: open path=%s exists=%s size=%s bytes",
-                   DB_PATH, DB_PATH.exists(), size)
-        try:
-            dblist = conn.execute("PRAGMA database_list").fetchall()
-            ulog.debug("db: PRAGMA database_list => %s", dblist)
-        except Exception as e:
-            ulog.debug("db: PRAGMA database_list failed: %s", e)
-        return conn
-    except Exception as e:
-        ulog.exception("db: connect failed for %s: %s", DB_PATH, e)
-        raise