Update scripts/backup_to_dataset.py

scripts/backup_to_dataset.py

@@ -4,34 +4,30 @@ import sys
 import datetime
 from huggingface_hub import HfApi, login
 
-# ===== 配置区域 =====
-DB_PATH = "/app/server/data/freeapi.db"          # 数据库文件路径
-DATASET_REPO = "lydgs/freellm-backup"         # 替换成你的私有数据集
-BACKUP_PREFIX = "freellm_backup"                  # 备份文件前缀
-RETENTION_DAYS = 30                               # 保留天数（超过则删除）
-# ===================
+# ===== 配置 =====
+DB_PATH = "/data/freeapi.db"                       # 数据库路径（与 DATABASE_URL 一致）
+DATASET_REPO = "lydgs/freellm-backup"          # 替换为你的数据集名称
+BACKUP_PREFIX = "freeapi_backup"                  # 备份文件前缀
+RETENTION_DAYS = 30                               # 保留天数
+# ================
 
 def backup_database():
-    # 检查数据库文件
     if not os.path.exists(DB_PATH):
         print(f"❌ 数据库文件不存在: {DB_PATH}")
         return False
 
-    # 生成备份文件名
-    timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
-    backup_name = f"{BACKUP_PREFIX}_{timestamp}.sqlite"
-
-    # 获取 Hugging Face Token
     token = os.getenv("HF_TOKEN")
     if not token:
         print("❌ 环境变量 HF_TOKEN 未设置")
         return False
 
-    # 登录
+    timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
+    backup_name = f"{BACKUP_PREFIX}_{timestamp}.db"
+
     login(token=token)
     api = HfApi()
 
-    # 1. 上传当前备份
+    # 上传当前备份
     try:
         api.upload_file(
             path_or_fileobj=DB_PATH,
@@ -39,33 +35,31 @@ def backup_database():
             repo_id=DATASET_REPO,
             repo_type="dataset"
         )
-        print(f"✅ 备份成功: https://huggingface.co/datasets/{DATASET_REPO}/blob/main/{backup_name}")
+        print(f"✅ 备份成功: {backup_name}")
     except Exception as e:
         print(f"❌ 上传失败: {e}")
         return False
 
-    # 2. 清理旧备份（保留最近 RETENTION_DAYS 天的文件）
+    # 清理旧备份（保留最近 RETENTION_DAYS 天）
     try:
         files = api.list_repo_files(repo_id=DATASET_REPO, repo_type="dataset")
         now = datetime.datetime.now()
-        deleted_count = 0
+        deleted = 0
         for f in files:
-            if not f.startswith(BACKUP_PREFIX) or not f.endswith(".sqlite"):
+            if not f.startswith(BACKUP_PREFIX) or not f.endswith(".db"):
                 continue
-            # 解析时间戳部分，文件名格式: prefix_YYYYMMDD_HHMMSS.sqlite
+            # 解析时间戳，文件名格式: freeapi_backup_YYYYMMDD_HHMMSS.db
             try:
-                parts = f.replace(".sqlite", "").split("_")
-                if len(parts) >= 3:
-                    ts_str = f"{parts[1]}_{parts[2]}"   # YYYYMMDD_HHMMSS
-                    file_time = datetime.datetime.strptime(ts_str, "%Y%m%d_%H%M%S")
-                    if (now - file_time).days > RETENTION_DAYS:
-                        api.delete_file(path_in_repo=f, repo_id=DATASET_REPO, repo_type="dataset")
-                        print(f"🗑️ 已删除旧备份: {f}")
-                        deleted_count += 1
-            except Exception as parse_err:
-                print(f"⚠️ 跳过无法解析的文件: {f}, 原因: {parse_err}")
-        if deleted_count:
-            print(f"✅ 清理完成，共删除 {deleted_count} 个旧备份")
+                ts_str = f.replace(BACKUP_PREFIX + "_", "").replace(".db", "")
+                file_time = datetime.datetime.strptime(ts_str, "%Y%m%d_%H%M%S")
+                if (now - file_time).days > RETENTION_DAYS:
+                    api.delete_file(path_in_repo=f, repo_id=DATASET_REPO, repo_type="dataset")
+                    print(f"🗑️ 已删除旧备份: {f}")
+                    deleted += 1
+            except Exception:
+                pass
+        if deleted:
+            print(f"✅ 清理完成，共删除 {deleted} 个旧备份")
         else:
             print("✅ 没有需要清理的旧备份")
     except Exception as e:
@@ -74,10 +68,10 @@ def backup_database():
     return True
 
 if __name__ == "__main__":
-    print(f"[{datetime.datetime.now()}] 开始备份并清理...")
+    print(f"[{datetime.datetime.now()}] 开始备份...")
     if backup_database():
-        print("任务完成")
+        print("备份任务完成")
         sys.exit(0)
     else:
-        print("任务失败")
+        print("备份任务失败")
         sys.exit(1)