sync_run.py

import os
import time
import subprocess
import threading
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

api = HfApi(token=TOKEN)

def download_data():
    """从 Dataset 下载数据到本地"""
    if not DATASET_ID:
        print("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}")

def upload_data():
    """定时上传数据到 Dataset"""
    while True:
        time.sleep(300)  # 每 5 分钟备份一次
        if DATASET_ID:
            try:
                api.upload_folder(
                    folder_path=LOCAL_DIR,
                    repo_id=DATASET_ID,
                    repo_type="dataset"
                )
                print("备份成功至 Dataset")
            except Exception as e:
                print(f"备份失败: {e}")

if __name__ == "__main__":
    # 1. 启动前先下载
    download_data()

    # 2. 启动定时备份线程
    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
    subprocess.run(["python", "webui.py", "--host", "0.0.0.0", "--port", "7860"])