utils.py 文件源码

def get_license_from_url(url):
    """Get the license abbreviation from an URL.

    Args:
        url(str): canonical url of the license.

    Returns:
        str: the corresponding license abbreviation.

    Raises:
        ValueError: when the url is not recognized
    """
    if not url:
        return

    split_url = urlsplit(url, scheme='http')

    if split_url.netloc.lower() == 'creativecommons.org':
        license = ['CC']
        match = _RE_LICENSE_URL.match(split_url.path)
        license.extend(part.upper() for part in match.groups() if part)
    elif split_url.netloc == 'arxiv.org':
        license = ['arXiv']
        match = _RE_LICENSE_URL.match(split_url.path)
        license.extend(part for part in match.groups() if part)
    else:
        raise ValueError('Unknown license URL')

    return u' '.join(license)