scripts/check_hf_storage.py

import os

from huggingface_hub import HfApi

from training_coach.storage import HuggingFaceDatasetHistoryStore


def main():
    repo_id = os.getenv("HF_HISTORY_DATASET_REPO", "").strip()
    token = os.getenv("HF_TOKEN")
    path_in_repo = os.getenv("HF_HISTORY_FILE", "completed_sessions.json")

    if not repo_id:
        raise SystemExit("Set HF_HISTORY_DATASET_REPO before running this check.")
    if not token:
        raise SystemExit("Set HF_TOKEN before running this check.")

    api = HfApi()
    repo_info = api.repo_info(repo_id=repo_id, repo_type="dataset", token=token)
    print(f"Dataset repo reachable: {repo_info.id}")

    store = HuggingFaceDatasetHistoryStore(
        repo_id=repo_id,
        token=token,
        path_in_repo=path_in_repo,
        api=api,
    )
    sessions = store.load_completed_sessions()
    print(f"History file: {path_in_repo}")
    print(f"Completed sessions loaded: {len(sessions)}")
    print("HF storage read check passed. No upload was performed.")


if __name__ == "__main__":
    main()