File size: 3,438 Bytes
84e9a1f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 | """Kaggle CPU kernel: Modal volume -> private HF dataset (and models -> private HF model repo). Needs 3 Kaggle secrets:
MODAL_TOKEN_ID, MODAL_TOKEN_SECRET, HF_TOKEN. Idempotent: a prefix already present on HF is skipped. Uploads per path."""
import json, os, shutil, subprocess, sys, time
from kaggle_secrets import UserSecretsClient
sec = UserSecretsClient()
os.environ["MODAL_TOKEN_ID"] = sec.get_secret("MODAL_TOKEN_ID")
os.environ["MODAL_TOKEN_SECRET"] = sec.get_secret("MODAL_TOKEN_SECRET")
HF = sec.get_secret("HF_TOKEN")
subprocess.run([sys.executable, "-m", "pip", "install", "-q", "modal"], check=True)
from huggingface_hub import HfApi
api = HfApi(token=HF)
assert api.whoami()["name"] == "Arko007", "wrong HF account"
DS, MD = "Arko007/Yoga-1M", "Arko007/asanaai-conference-runs"
# (volume path, repo, repo_type, path_in_repo) -- smallest / most valuable first
PATHS = [("cue", DS, "dataset", "vol/cue"), ("transcripts", DS, "dataset", "vol/transcripts"),
("labelsrc", DS, "dataset", "vol/labelsrc"), ("eval", DS, "dataset", "vol/eval"), ("folds", DS, "dataset", "vol/folds"),
("csv/cue", DS, "dataset", "vol/csv/cue"), ("photos", DS, "dataset", "vol/photos"),
("runs", MD, "model", "runs"),
("csv/cue2", DS, "dataset", "vol/csv/cue2"), ("landmarks", DS, "dataset", "vol/landmarks"),
("landmarks_new", DS, "dataset", "vol/landmarks_new"),
("stgcn/cueT2_f0", DS, "dataset", "vol/stgcn/cueT2_f0"), ("stgcn/cueT2_f1", DS, "dataset", "vol/stgcn/cueT2_f1"),
("stgcn/cueT2_f2", DS, "dataset", "vol/stgcn/cueT2_f2"), ("stgcn/cueT2_all", DS, "dataset", "vol/stgcn/cueT2_all")]
report = []
for vp, repo, rt, dst in PATHS:
t0 = time.time()
try:
have = [x for x in api.list_repo_tree(repo, path_in_repo=dst, repo_type=rt)] if True else []
except Exception:
have = []
if have:
report.append({"path": vp, "status": "already_on_hf"}); print("skip (already on HF):", vp, flush=True); continue
tmp = f"/kaggle/working/stage/{vp.replace('/', '__')}"
shutil.rmtree(tmp, ignore_errors=True); os.makedirs(tmp)
r = subprocess.run(["modal", "volume", "get", "asanaai-data", vp, tmp, "--force"], capture_output=True, text=True)
if r.returncode != 0:
report.append({"path": vp, "status": "modal_get_failed", "err": (r.stderr or r.stdout)[-300:]}); print("FAILED get", vp, r.stderr[-200:], flush=True); continue
base = os.path.join(tmp, os.path.basename(vp))
src = base if os.path.exists(base) else tmp
size = sum(os.path.getsize(os.path.join(d, f)) for d, _, fs in os.walk(src) for f in fs) if os.path.isdir(src) else os.path.getsize(src)
if os.path.isdir(src):
api.upload_folder(folder_path=src, path_in_repo=dst, repo_id=repo, repo_type=rt, commit_message=f"transfer {vp}")
else:
api.upload_file(path_or_fileobj=src, path_in_repo=dst, repo_id=repo, repo_type=rt, commit_message=f"transfer {vp}")
shutil.rmtree(tmp, ignore_errors=True)
report.append({"path": vp, "status": "ok", "mb": round(size / 1e6, 1), "secs": round(time.time() - t0)})
print("OK", vp, f"{size/1e6:.0f} MB", f"{time.time()-t0:.0f}s", flush=True)
json.dump(report, open("/kaggle/working/transfer_report.json", "w"), indent=1)
api.upload_file(path_or_fileobj="/kaggle/working/transfer_report.json", path_in_repo="transfer_report.json", repo_id=DS, repo_type="dataset")
print(json.dumps(report, indent=1))
|