Create sync_data.sh
Browse files- sync_data.sh +115 -0
sync_data.sh
ADDED
|
@@ -0,0 +1,115 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/bin/sh
|
| 2 |
+
|
| 3 |
+
# 检查环境变量
|
| 4 |
+
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
|
| 5 |
+
echo "缺少必要的环境变量 HF_TOKEN 或 DATASET_ID"
|
| 6 |
+
exit 1
|
| 7 |
+
fi
|
| 8 |
+
|
| 9 |
+
# 上传备份
|
| 10 |
+
cat > /tmp/hf_sync.py << 'EOL'
|
| 11 |
+
from huggingface_hub import HfApi
|
| 12 |
+
import sys
|
| 13 |
+
import os
|
| 14 |
+
|
| 15 |
+
def manage_backups(api, repo_id, max_files=50):
|
| 16 |
+
files = api.list_repo_files(repo_id=repo_id, repo_type="dataset")
|
| 17 |
+
backup_files = [f for f in files if f.startswith('webui_backup_') and f.endswith('.db')]
|
| 18 |
+
backup_files.sort()
|
| 19 |
+
|
| 20 |
+
if len(backup_files) >= max_files:
|
| 21 |
+
files_to_delete = backup_files[:(len(backup_files) - max_files + 1)]
|
| 22 |
+
for file_to_delete in files_to_delete:
|
| 23 |
+
try:
|
| 24 |
+
api.delete_file(path_in_repo=file_to_delete, repo_id=repo_id, repo_type="dataset")
|
| 25 |
+
print(f'已删除旧备份: {file_to_delete}')
|
| 26 |
+
except Exception as e:
|
| 27 |
+
print(f'删除 {file_to_delete} 时出错: {str(e)}')
|
| 28 |
+
|
| 29 |
+
def upload_backup(file_path, file_name, token, repo_id):
|
| 30 |
+
api = HfApi(token=token)
|
| 31 |
+
try:
|
| 32 |
+
api.upload_file(
|
| 33 |
+
path_or_fileobj=file_path,
|
| 34 |
+
path_in_repo=file_name,
|
| 35 |
+
repo_id=repo_id,
|
| 36 |
+
repo_type="dataset"
|
| 37 |
+
)
|
| 38 |
+
print(f"成功上传 {file_name}")
|
| 39 |
+
|
| 40 |
+
manage_backups(api, repo_id)
|
| 41 |
+
except Exception as e:
|
| 42 |
+
print(f"文件上传出错: {str(e)}")
|
| 43 |
+
|
| 44 |
+
# 下载最新备份
|
| 45 |
+
def download_latest_backup(token, repo_id):
|
| 46 |
+
try:
|
| 47 |
+
api = HfApi(token=token)
|
| 48 |
+
files = api.list_repo_files(repo_id=repo_id, repo_type="dataset")
|
| 49 |
+
backup_files = [f for f in files if f.startswith('webui_backup_') and f.endswith('.db')]
|
| 50 |
+
|
| 51 |
+
if not backup_files:
|
| 52 |
+
print("未找到备份文件")
|
| 53 |
+
return
|
| 54 |
+
|
| 55 |
+
latest_backup = sorted(backup_files)[-1]
|
| 56 |
+
|
| 57 |
+
filepath = api.hf_hub_download(
|
| 58 |
+
repo_id=repo_id,
|
| 59 |
+
filename=latest_backup,
|
| 60 |
+
repo_type="dataset"
|
| 61 |
+
)
|
| 62 |
+
|
| 63 |
+
if filepath and os.path.exists(filepath):
|
| 64 |
+
os.makedirs('./data', exist_ok=True)
|
| 65 |
+
os.system(f'cp "{filepath}" ./data/webui.db')
|
| 66 |
+
print(f"成功从 {latest_backup} 恢复备份")
|
| 67 |
+
|
| 68 |
+
except Exception as e:
|
| 69 |
+
print(f"下载备份时出错: {str(e)}")
|
| 70 |
+
|
| 71 |
+
if __name__ == "__main__":
|
| 72 |
+
action = sys.argv[1]
|
| 73 |
+
token = sys.argv[2]
|
| 74 |
+
repo_id = sys.argv[3]
|
| 75 |
+
|
| 76 |
+
if action == "upload":
|
| 77 |
+
file_path = sys.argv[4]
|
| 78 |
+
file_name = sys.argv[5]
|
| 79 |
+
upload_backup(file_path, file_name, token, repo_id)
|
| 80 |
+
elif action == "download":
|
| 81 |
+
download_latest_backup(token, repo_id)
|
| 82 |
+
EOL
|
| 83 |
+
|
| 84 |
+
# 首次启动时下载最新备份
|
| 85 |
+
echo "正在从 HuggingFace 下载最新备份..."
|
| 86 |
+
python3 /tmp/hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}"
|
| 87 |
+
|
| 88 |
+
# 同步函数
|
| 89 |
+
sync_data() {
|
| 90 |
+
while true; do
|
| 91 |
+
echo "开始同步进程 $(date)"
|
| 92 |
+
|
| 93 |
+
if [ -f "./data/webui.db" ]; then
|
| 94 |
+
timestamp=$(date +%Y%m%d_%H%M%S)
|
| 95 |
+
backup_file="webui_backup_${timestamp}.db"
|
| 96 |
+
|
| 97 |
+
# 复制数据库文件
|
| 98 |
+
cp ./data/webui.db "/tmp/${backup_file}"
|
| 99 |
+
|
| 100 |
+
echo "正在上传备份到 HuggingFace..."
|
| 101 |
+
python3 /tmp/hf_sync.py upload "${HF_TOKEN}" "${DATASET_ID}" "/tmp/${backup_file}" "${backup_file}"
|
| 102 |
+
|
| 103 |
+
rm -f "/tmp/${backup_file}"
|
| 104 |
+
else
|
| 105 |
+
echo "数据库文件不存在,等待下次同步..."
|
| 106 |
+
fi
|
| 107 |
+
|
| 108 |
+
SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
|
| 109 |
+
echo "下次同步将在 ${SYNC_INTERVAL} 秒后进行..."
|
| 110 |
+
sleep $SYNC_INTERVAL
|
| 111 |
+
done
|
| 112 |
+
}
|
| 113 |
+
|
| 114 |
+
# 后台启动同步进程
|
| 115 |
+
sync_data &
|