def compute_modalities_response()

in services/worker/src/worker/job_runners/dataset/modalities.py [0:0]

• 29 lines of code
• 7 McCabe index (conditional complexity)

def compute_modalities_response(dataset: str) -> DatasetModalitiesResponse:
    """
    Get the response of 'dataset-modalities' for one specific dataset on huggingface.co.

    Args:
        dataset (`str`):
            A namespace (user or an organization) and a repo name separated by a `/`.

    Raises:
        [~`libcommon.exceptions.PreviousStepFormatError`]:
            If the content of the previous step has not the expected format

    Returns:
        `tuple[DatasetModalitiesResponse, float]`: An object with the modalities_response and the progress.
    """
    logging.info(f"compute 'dataset-modalities' for {dataset=}")

    modalities: set[DatasetModality] = set()
    try:
        modalities.update(detect_modalities_from_features(dataset))
    except PreviousStepFormatError:
        raise
    except Exception:
        logging.info(f"failed to detect modalities from features of {dataset=}")
        pass

    try:
        modalities.update(detect_modalities_from_url_columns(dataset))
    except PreviousStepFormatError:
        raise
    except Exception:
        logging.info(f"failed to detect modalities from file types of {dataset=}")
        pass

    try:
        modalities.update(detect_modalities_from_filetypes(dataset))
    except PreviousStepFormatError:
        raise
    except Exception:
        logging.info(f"failed to detect modalities from file types of {dataset=}")
        pass

    return DatasetModalitiesResponse(
        {
            "modalities": sorted(modalities),
        }
    )