Chuyển từ dữ liệu tạm sang runtime để dữ liệu không bị mất

backend/auth.py

@@ -11,6 +11,7 @@ import sqlite3
 from datetime import datetime, timedelta
 from backend.admin_config import is_bootstrap_admin_email
 from backend.runtime_paths import DB_PATH
+from backend.db_sync import schedule_sync as _schedule_sync
 
 
 def _get_connection():
@@ -80,6 +81,7 @@ def register_user(email: str, password: str, display_name: str = None) -> dict:
             "role": role,
         }
         print(f"[AUTH] ✅ Đăng ký thành công: {email}")
+        _schedule_sync()
         return {"success": True, "message": "Đăng ký thành công!", "user": user}
 
     except Exception as e:
@@ -209,6 +211,7 @@ def reset_password(email: str, token: str, new_password: str) -> dict:
         conn.commit()
 
         print(f"[AUTH] ✅ Password reset for: {email}")
+        _schedule_sync()
         return {"success": True, "message": "Đặt lại mật khẩu thành công!"}
 
     finally:

backend/db.py

@@ -9,6 +9,7 @@ import time
 from datetime import datetime
 from backend.admin_config import get_admin_emails
 from backend.runtime_paths import DB_PATH
+from backend.db_sync import schedule_sync as _schedule_sync
 
 
 def _get_connection():
@@ -249,6 +250,7 @@ def save_conversation(ma_cuoc_tro_chuyen: str, tieu_de: str = "Cuộc trò chuy
         conn.commit()
     finally:
         conn.close()
+    _schedule_sync()
 
 
 def save_message(ma_cuoc_tro_chuyen: str, vai_tro: str, noi_dung: str,
@@ -268,9 +270,11 @@ def save_message(ma_cuoc_tro_chuyen: str, vai_tro: str, noi_dung: str,
             ),
         )
         conn.commit()
-        return cursor.lastrowid
+        lastrowid = cursor.lastrowid
     finally:
         conn.close()
+    _schedule_sync()
+    return lastrowid
 
 
 def save_message_sources(ma_tin_nhan: int, sources: list):
@@ -457,6 +461,7 @@ def delete_conversation(ma_cuoc_tro_chuyen: str):
         print(f"[DB] 🗑️ Deleted conversation: {ma_cuoc_tro_chuyen}")
     finally:
         conn.close()
+    _schedule_sync()
 
 
 # ======================== TÀI LIỆU (GỘP user + admin) ========================
@@ -532,6 +537,7 @@ def save_phan_hoi(ma_tin_nhan: int, loai: str, ma_nguoi_dung: str = None,
         conn.commit()
     finally:
         conn.close()
+    _schedule_sync()
 
 
 def load_phan_hoi_by_tin_nhan(ma_tin_nhan: int) -> list:
@@ -601,9 +607,10 @@ def update_user_role(ma_nguoi_dung: str, vai_tro: str) -> bool:
             (vai_tro, ma_nguoi_dung),
         )
         conn.commit()
-        return True
     finally:
         conn.close()
+    _schedule_sync()
+    return True
 
 
 def lock_user(ma_nguoi_dung: str) -> bool:
@@ -615,9 +622,10 @@ def lock_user(ma_nguoi_dung: str) -> bool:
             (ma_nguoi_dung,),
         )
         conn.commit()
-        return True
     finally:
         conn.close()
+    _schedule_sync()
+    return True
 
 
 def unlock_user(ma_nguoi_dung: str) -> bool:
@@ -629,9 +637,10 @@ def unlock_user(ma_nguoi_dung: str) -> bool:
             (ma_nguoi_dung,),
         )
         conn.commit()
-        return True
     finally:
         conn.close()
+    _schedule_sync()
+    return True
 
 
 # ======================== ADMIN: LỊCH SỬ HỎI ĐÁP TOÀN HỆ THỐNG ========================

backend/db_sync.py

@@ -0,0 +1,113 @@
+"""
+Đồng bộ chatbot.db lên Hugging Face Dataset repo để dữ liệu bền vững.
+
+Chỉ hoạt động khi chạy trên HF Space và có cấu hình:
+  - HF_DATASET_REPO
+  - HF_TOKEN
+"""
+
+from __future__ import annotations
+
+import os
+import shutil
+import sqlite3
+import threading
+import time
+
+_lock = threading.Lock()
+_last_sync: float = 0
+_MIN_INTERVAL = 30  # tối thiểu 30 giây giữa 2 lần sync
+
+
+def _get_config() -> tuple[str, str, str] | None:
+    repo = os.getenv("HF_DATASET_REPO", "").strip()
+    token = os.getenv("HF_TOKEN") or os.getenv("HUGGINGFACE_TOKEN")
+    if not repo or not token:
+        return None
+    revision = os.getenv("HF_DATASET_REVISION", "main").strip() or "main"
+    return repo, token, revision
+
+
+def _safe_copy_db(db_path: str, dest: str) -> bool:
+    """Tạo bản sao sạch của SQLite (checkpoint WAL trước khi copy)."""
+    try:
+        conn = sqlite3.connect(db_path)
+        conn.execute("PRAGMA wal_checkpoint(TRUNCATE)")
+        conn.close()
+    except Exception:
+        pass
+
+    try:
+        shutil.copy2(db_path, dest)
+        return True
+    except Exception as e:
+        print(f"[DB_SYNC] Copy failed: {e}")
+        return False
+
+
+def _do_upload(db_path: str, repo: str, token: str, revision: str) -> bool:
+    """Upload chatbot.db lên dataset repo."""
+    try:
+        from huggingface_hub import HfApi
+        api = HfApi(token=token)
+
+        tmp_path = db_path + ".sync_copy"
+        if not _safe_copy_db(db_path, tmp_path):
+            return False
+
+        api.upload_file(
+            path_or_fileobj=tmp_path,
+            path_in_repo="chatbot.db",
+            repo_id=repo,
+            repo_type="dataset",
+            revision=revision,
+            commit_message="Auto-sync chatbot.db",
+        )
+
+        try:
+            os.remove(tmp_path)
+        except OSError:
+            pass
+
+        print(f"[DB_SYNC] Uploaded chatbot.db to {repo}")
+        return True
+
+    except Exception as e:
+        print(f"[DB_SYNC] Upload failed: {e}")
+        return False
+
+
+def schedule_sync(db_path: str | None = None) -> None:
+    """
+    Lên lịch đồng bộ DB lên dataset repo (chạy background, có rate-limit).
+    Gọi hàm này sau mỗi lần ghi quan trọng vào DB.
+    """
+    global _last_sync
+
+    config = _get_config()
+    if not config:
+        return
+
+    now = time.time()
+    if now - _last_sync < _MIN_INTERVAL:
+        return
+
+    if db_path is None:
+        from backend.runtime_paths import DB_PATH
+        db_path = DB_PATH
+
+    if not os.path.exists(db_path):
+        return
+
+    repo, token, revision = config
+
+    def _sync():
+        global _last_sync
+        with _lock:
+            if time.time() - _last_sync < _MIN_INTERVAL:
+                return
+            _last_sync = time.time()
+            _do_upload(db_path, repo, token, revision)
+
+    thread = threading.Thread(target=_sync, daemon=True)
+    thread.start()

scripts/bootstrap_space_data.py

@@ -59,6 +59,13 @@ def _copy_file_if_missing(src: str, dst: str) -> None:
         shutil.copy2(src, dst)
 
 
+def _copy_file_always(src: str, dst: str) -> None:
+    """Luôn copy file từ src sang dst (ghi đè nếu đã tồn tại)."""
+    if os.path.exists(src):
+        os.makedirs(os.path.dirname(dst), exist_ok=True)
+        shutil.copy2(src, dst)
+
+
 def _vector_dir_ready() -> bool:
     return os.path.isdir(VECTOR_DIR) and any(Path(VECTOR_DIR).iterdir())
 
@@ -131,7 +138,8 @@ def bootstrap_space_data(force: bool = False) -> bool:
             _copy_tree(pdf_src, PDF_DIR)
 
         db_src = os.path.join(snapshot_dir, "chatbot.db")
-        _copy_file_if_missing(db_src, DB_PATH)
+        _copy_file_always(db_src, DB_PATH)
+        print(f"[BOOTSTRAP] Đã tải chatbot.db mới nhất từ dataset repo")
 
     meta = {
         "repo_id": repo_id,