hf_video_storage.py

"""
hf_video_storage.py — lazy-bubble large file storage backed by a private
HF dataset repo. Upload returns immediately; the recipient's download
pulls the bytes then permanently deletes the file from the repo.
"""

import os
import uuid
import requests
from huggingface_hub import HfApi

VIDEO_REPO = os.environ.get("VIDEO_DATASET_REPO", "Brighton233j/Video-Temp-Storage")
_HF_TOKEN = os.environ.get("HF_TOKEN") or os.environ.get("Hf_token")
_api = HfApi(token=_HF_TOKEN)


def upload_large_file(file_bytes, filename, mime_type=None):
    ext = os.path.splitext(filename)[1] or ""
    path_in_repo = f"{uuid.uuid4().hex}{ext}"

    _api.upload_file(
        path_or_fileobj=file_bytes,
        path_in_repo=path_in_repo,
        repo_id=VIDEO_REPO,
        repo_type="dataset",
        commit_message=f"Upload {filename}",
    )

    return {"id": path_in_repo, "name": filename}


def download_and_delete(file_id):
    url = f"https://huggingface.co/datasets/{VIDEO_REPO}/resolve/main/{file_id}"
    resp = requests.get(url, headers={"Authorization": f"Bearer {_HF_TOKEN}"})
    resp.raise_for_status()
    file_bytes = resp.content

    _api.delete_file(
        path_in_repo=file_id,
        repo_id=VIDEO_REPO,
        repo_type="dataset",
        commit_message=f"Delete downloaded file {file_id}",
    )

    return file_bytes, file_id