Update sync_run.py

sync_run.py

@@ -7,43 +7,62 @@ from huggingface_hub import HfApi, snapshot_download
 # 配置
 TOKEN = os.getenv("HF_TOKEN")
 DATASET_ID = os.getenv("DATASET_ID")
-LOCAL_DIR = "data"  # 项目存储账号和数据库的目录
+# 需要备份的文件夹列表
+SYNC_FOLDERS = ["data", "output"] 
 
 api = HfApi(token=TOKEN)
 
 def download_data():
-    """从 Dataset 下载数据到本地"""
+    """初始化：从 Dataset 下载所有备份文件"""
     if not DATASET_ID:
-        print("未配置 DATASET_ID，跳过下载")
+        print("[System] 未配置 DATASET_ID，跳过下载。")
         return
-    try:
-        print(f"正在从 {DATASET_ID} 同步数据...")
-        snapshot_download(
-            repo_id=DATASET_ID,
-            repo_type="dataset",
-            local_dir=LOCAL_DIR,
-            token=TOKEN
-        )
-        print("同步完成")
-    except Exception as e:
-        print(f"下载失败（可能是首次运行，仓库为空）: {e}")
+    for folder in SYNC_FOLDERS:
+        try:
+            print(f"[System] 正在同步云端 {folder} 文件夹...")
+            snapshot_download(
+                repo_id=DATASET_ID,
+                repo_type="dataset",
+                local_dir=folder,
+                allow_patterns=f"{folder}/*", # 只下载对应文件夹内容
+                token=TOKEN
+            )
+        except Exception as e:
+            print(f"[System] 下载 {folder} 失败 (可能仓库为空): {e}")
 
 def upload_data():
-    """定时上传数据到 Dataset"""
+    """定时任务：上传本地修改到 Dataset"""
     while True:
-        time.sleep(300)  # 每 5 分钟备份一次
-        if DATASET_ID:
+        time.sleep(300) # 每 5 分钟检查一次
+        if not DATASET_ID:
+            continue
+            
+        print(f"[Backup] {time.strftime('%Y-%m-%d %H:%M:%S')} 开始检查文件变动...")
+        for folder in SYNC_FOLDERS:
+            if not os.path.exists(folder) or not os.listdir(folder):
+                continue
+                
             try:
+                # upload_folder 会自动比对哈希值，只有文件变了才会真的上传
                 api.upload_folder(
-                    folder_path=LOCAL_DIR,
+                    folder_path=folder,
+                    path_in_repo=folder, # 在 Dataset 中对应的路径
                     repo_id=DATASET_ID,
-                    repo_type="dataset"
+                    repo_type="dataset",
+                    commit_message=f"Scheduled backup of {folder}",
+                    run_as_future=True # 异步运行，不阻塞
                 )
-                print("备份成功至 Dataset")
             except Exception as e:
-                print(f"备份失败: {e}")
+                # 忽略 "No files have been modified" 这种不算错误的错误
+                if "No files have been modified" not in str(e):
+                    print(f"[Backup] {folder} 上传失败: {e}")
+        print(f"[Backup] 检查完毕。")
 
 if __name__ == "__main__":
+    # 创建本地文件夹防止报错
+    for f in SYNC_FOLDERS:
+        os.makedirs(f, exist_ok=True)
+
     # 1. 启动前先下载
     download_data()
 
@@ -51,6 +70,6 @@ if __name__ == "__main__":
     backup_thread = threading.Thread(target=upload_data, daemon=True)
     backup_thread.start()
 
-    # 3. 运行原有的 网页UI (注意：此处要和你的 Dockerfile 启动命令一致)
-    # 使用 subprocess 运行，防止主进程阻塞
+    # 3. 运行 网页UI
+    print("[System] 正在启动 网页UI...")
     subprocess.run(["python", "webui.py", "--host", "0.0.0.0", "--port", "7860"])