Create scripts/sync_hf.py
Browse files- scripts/sync_hf.py +149 -0
scripts/sync_hf.py
ADDED
|
@@ -0,0 +1,149 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
"""
|
| 3 |
+
VS Code Space β HF Dataset Persistence
|
| 4 |
+
Restores /data on boot, auto-saves every 5 minutes
|
| 5 |
+
"""
|
| 6 |
+
|
| 7 |
+
import os, sys, time, threading, shutil, traceback
|
| 8 |
+
from pathlib import Path
|
| 9 |
+
from datetime import datetime
|
| 10 |
+
from huggingface_hub import HfApi, snapshot_download
|
| 11 |
+
|
| 12 |
+
# ββ Config βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 13 |
+
|
| 14 |
+
HF_TOKEN = os.environ.get("HF_TOKEN", "")
|
| 15 |
+
DATASET_REPO = os.environ.get("REPO", "") # e.g. abc1181/vscode-storage
|
| 16 |
+
SYNC_INTERVAL = int(os.environ.get("SYNC_INTERVAL", "300")) # 5 mins default
|
| 17 |
+
DATA_DIR = Path("/data")
|
| 18 |
+
PATH_IN_REPO = "workspace" # folder name inside dataset repo
|
| 19 |
+
|
| 20 |
+
IGNORE = [
|
| 21 |
+
"*.log", "*.lock", "*.tmp", "*.pid",
|
| 22 |
+
"__pycache__", "node_modules/**",
|
| 23 |
+
".git/**", "*.pyc"
|
| 24 |
+
]
|
| 25 |
+
|
| 26 |
+
# ββ Setup ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 27 |
+
|
| 28 |
+
DATA_DIR.mkdir(parents=True, exist_ok=True)
|
| 29 |
+
|
| 30 |
+
if not HF_TOKEN:
|
| 31 |
+
print("[SYNC] WARNING: HF_TOKEN not set β persistence disabled")
|
| 32 |
+
sys.exit(0)
|
| 33 |
+
|
| 34 |
+
if not DATASET_REPO:
|
| 35 |
+
print("[SYNC] WARNING: REPO not set β persistence disabled")
|
| 36 |
+
sys.exit(0)
|
| 37 |
+
|
| 38 |
+
api = HfApi(token=HF_TOKEN)
|
| 39 |
+
|
| 40 |
+
# ββ Ensure dataset repo exists βββββββββββββββββββββββββββββββββββββββββββββββ
|
| 41 |
+
|
| 42 |
+
def ensure_repo():
|
| 43 |
+
try:
|
| 44 |
+
api.repo_info(repo_id=DATASET_REPO, repo_type="dataset")
|
| 45 |
+
print(f"[SYNC] Dataset found: {DATASET_REPO}")
|
| 46 |
+
return True
|
| 47 |
+
except Exception:
|
| 48 |
+
try:
|
| 49 |
+
api.create_repo(repo_id=DATASET_REPO, repo_type="dataset", private=True)
|
| 50 |
+
print(f"[SYNC] Created dataset: {DATASET_REPO}")
|
| 51 |
+
return True
|
| 52 |
+
except Exception as e:
|
| 53 |
+
print(f"[SYNC] Failed to find/create dataset: {e}")
|
| 54 |
+
return False
|
| 55 |
+
|
| 56 |
+
# ββ Restore /data from dataset on boot βββββββββββββββββββββββββββββββββββββββ
|
| 57 |
+
|
| 58 |
+
def restore():
|
| 59 |
+
print(f"[SYNC] Restoring /data from {DATASET_REPO}...")
|
| 60 |
+
try:
|
| 61 |
+
files = list(api.list_repo_files(repo_id=DATASET_REPO, repo_type="dataset"))
|
| 62 |
+
ws_files = [f for f in files if f.startswith(f"{PATH_IN_REPO}/")]
|
| 63 |
+
|
| 64 |
+
if not ws_files:
|
| 65 |
+
print("[SYNC] No files in dataset yet β starting fresh")
|
| 66 |
+
return
|
| 67 |
+
|
| 68 |
+
print(f"[SYNC] Found {len(ws_files)} files β downloading...")
|
| 69 |
+
import tempfile
|
| 70 |
+
with tempfile.TemporaryDirectory() as tmpdir:
|
| 71 |
+
snapshot_download(
|
| 72 |
+
repo_id=DATASET_REPO,
|
| 73 |
+
repo_type="dataset",
|
| 74 |
+
allow_patterns=f"{PATH_IN_REPO}/**",
|
| 75 |
+
local_dir=tmpdir,
|
| 76 |
+
token=HF_TOKEN,
|
| 77 |
+
)
|
| 78 |
+
src = Path(tmpdir) / PATH_IN_REPO
|
| 79 |
+
if src.exists():
|
| 80 |
+
for item in src.rglob("*"):
|
| 81 |
+
if item.is_file():
|
| 82 |
+
dest = DATA_DIR / item.relative_to(src)
|
| 83 |
+
dest.parent.mkdir(parents=True, exist_ok=True)
|
| 84 |
+
shutil.copy2(str(item), str(dest))
|
| 85 |
+
print("[SYNC] β
Restore complete!")
|
| 86 |
+
|
| 87 |
+
except Exception as e:
|
| 88 |
+
print(f"[SYNC] Restore failed: {e}")
|
| 89 |
+
traceback.print_exc()
|
| 90 |
+
|
| 91 |
+
# ββ Save /data to dataset βββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 92 |
+
|
| 93 |
+
def save():
|
| 94 |
+
try:
|
| 95 |
+
file_count = sum(1 for _, _, fs in os.walk(DATA_DIR) for _ in fs)
|
| 96 |
+
if file_count == 0:
|
| 97 |
+
print("[SYNC] Nothing to save β /data is empty")
|
| 98 |
+
return
|
| 99 |
+
|
| 100 |
+
print(f"[SYNC] Uploading {file_count} files β {DATASET_REPO}...")
|
| 101 |
+
api.upload_folder(
|
| 102 |
+
folder_path=str(DATA_DIR),
|
| 103 |
+
path_in_repo=PATH_IN_REPO,
|
| 104 |
+
repo_id=DATASET_REPO,
|
| 105 |
+
repo_type="dataset",
|
| 106 |
+
token=HF_TOKEN,
|
| 107 |
+
commit_message=f"autosave {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}",
|
| 108 |
+
ignore_patterns=IGNORE,
|
| 109 |
+
)
|
| 110 |
+
print(f"[SYNC] πΎ Saved at {datetime.now().strftime('%H:%M:%S')}")
|
| 111 |
+
|
| 112 |
+
except Exception as e:
|
| 113 |
+
print(f"[SYNC] Save failed: {e}")
|
| 114 |
+
traceback.print_exc()
|
| 115 |
+
|
| 116 |
+
# ββ Background sync loop ββββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 117 |
+
|
| 118 |
+
def sync_loop(stop_event):
|
| 119 |
+
print(f"[SYNC] Auto-save loop started (every {SYNC_INTERVAL}s)")
|
| 120 |
+
while not stop_event.is_set():
|
| 121 |
+
if stop_event.wait(timeout=SYNC_INTERVAL):
|
| 122 |
+
break
|
| 123 |
+
print(f"[SYNC] Periodic save at {datetime.now().isoformat()}")
|
| 124 |
+
save()
|
| 125 |
+
|
| 126 |
+
# ββ Main ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 127 |
+
|
| 128 |
+
def main():
|
| 129 |
+
if not ensure_repo():
|
| 130 |
+
sys.exit(1)
|
| 131 |
+
|
| 132 |
+
restore()
|
| 133 |
+
|
| 134 |
+
stop_event = threading.Event()
|
| 135 |
+
t = threading.Thread(target=sync_loop, args=(stop_event,), daemon=True)
|
| 136 |
+
t.start()
|
| 137 |
+
|
| 138 |
+
# Keep alive β VS Code server runs separately
|
| 139 |
+
try:
|
| 140 |
+
while True:
|
| 141 |
+
time.sleep(60)
|
| 142 |
+
except KeyboardInterrupt:
|
| 143 |
+
print("[SYNC] Shutting down β final save...")
|
| 144 |
+
stop_event.set()
|
| 145 |
+
save()
|
| 146 |
+
print("[SYNC] Done.")
|
| 147 |
+
|
| 148 |
+
if __name__ == "__main__":
|
| 149 |
+
main()
|