app/services/hf_service.py

from huggingface_hub import HfApi
from huggingface_hub.errors import HfHubHTTPError
from tenacity import retry, stop_after_attempt, wait_exponential
from app.utils.validation import safe_dataset_id
from app.services.logger import get_logger

log = get_logger(__name__)
api = HfApi()

@retry(stop=stop_after_attempt(3), wait=wait_exponential(min=1, max=8))
def dataset_exists(dataset_name: str) -> bool:
    safe_dataset_id(dataset_name)
    try:
        api.dataset_info(dataset_name)
    except HfHubHTTPError as e:
        log.error(f"Hugging Face dataset lookup failed: {e}")
        raise
    return True