Upload sync_data.py
Browse files- sync_data.py +83 -0
sync_data.py
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import shutil
|
| 3 |
+
import time
|
| 4 |
+
import threading
|
| 5 |
+
from pathlib import Path
|
| 6 |
+
from datetime import datetime
|
| 7 |
+
from huggingface_hub import HfApi, hf_hub_download, list_repo_files
|
| 8 |
+
|
| 9 |
+
DATA_DIR = Path(os.environ.get("DATA_DIR", "/root/.openclaw"))
|
| 10 |
+
DATASET_REPO_ID = os.environ.get("DATASET_REPO_ID", "your-username/your-dataset")
|
| 11 |
+
SYNC_INTERVAL = int(os.environ.get("SYNC_INTERVAL", 300))
|
| 12 |
+
ARCHIVE_NAME = "openclaw_backup.tar.gz"
|
| 13 |
+
|
| 14 |
+
api = HfApi()
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
def download_from_dataset():
|
| 18 |
+
DATA_DIR.mkdir(parents=True, exist_ok=True)
|
| 19 |
+
try:
|
| 20 |
+
files = list_repo_files(repo_id=DATASET_REPO_ID, repo_type="dataset")
|
| 21 |
+
if ARCHIVE_NAME in files:
|
| 22 |
+
print(f"[{datetime.now()}] Downloading {ARCHIVE_NAME}...")
|
| 23 |
+
archive_path = hf_hub_download(
|
| 24 |
+
repo_id=DATASET_REPO_ID,
|
| 25 |
+
filename=ARCHIVE_NAME,
|
| 26 |
+
repo_type="dataset",
|
| 27 |
+
local_dir="/tmp"
|
| 28 |
+
)
|
| 29 |
+
shutil.unpack_archive(archive_path, DATA_DIR)
|
| 30 |
+
os.remove(archive_path)
|
| 31 |
+
print(f"[{datetime.now()}] Data restored to {DATA_DIR}")
|
| 32 |
+
else:
|
| 33 |
+
print(f"[{datetime.now()}] No backup found, starting fresh")
|
| 34 |
+
except Exception as e:
|
| 35 |
+
print(f"[{datetime.now()}] Download failed: {e}")
|
| 36 |
+
|
| 37 |
+
|
| 38 |
+
def upload_to_dataset():
|
| 39 |
+
if not any(DATA_DIR.iterdir()):
|
| 40 |
+
print(f"[{datetime.now()}] No files to upload")
|
| 41 |
+
return
|
| 42 |
+
try:
|
| 43 |
+
archive_path = Path("/tmp") / ARCHIVE_NAME
|
| 44 |
+
shutil.make_archive(
|
| 45 |
+
archive_path.with_suffix("").with_suffix(""),
|
| 46 |
+
"gztar",
|
| 47 |
+
DATA_DIR
|
| 48 |
+
)
|
| 49 |
+
print(f"[{datetime.now()}] Uploading {ARCHIVE_NAME}...")
|
| 50 |
+
api.upload_file(
|
| 51 |
+
path_or_fileobj=str(archive_path),
|
| 52 |
+
path_in_repo=ARCHIVE_NAME,
|
| 53 |
+
repo_id=DATASET_REPO_ID,
|
| 54 |
+
repo_type="dataset",
|
| 55 |
+
)
|
| 56 |
+
os.remove(archive_path)
|
| 57 |
+
print(f"[{datetime.now()}] Upload completed")
|
| 58 |
+
except Exception as e:
|
| 59 |
+
print(f"[{datetime.now()}] Upload failed: {e}")
|
| 60 |
+
|
| 61 |
+
|
| 62 |
+
def sync_loop():
|
| 63 |
+
while True:
|
| 64 |
+
time.sleep(SYNC_INTERVAL)
|
| 65 |
+
upload_to_dataset()
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
def main():
|
| 69 |
+
print(f"Dataset: {DATASET_REPO_ID}")
|
| 70 |
+
print(f"Data dir: {DATA_DIR}")
|
| 71 |
+
print(f"Sync interval: {SYNC_INTERVAL}s")
|
| 72 |
+
|
| 73 |
+
download_from_dataset()
|
| 74 |
+
|
| 75 |
+
sync_thread = threading.Thread(target=sync_loop, daemon=True)
|
| 76 |
+
sync_thread.start()
|
| 77 |
+
|
| 78 |
+
while True:
|
| 79 |
+
time.sleep(3600)
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
if __name__ == "__main__":
|
| 83 |
+
main()
|