Make NbGitPuller URL for input URL

nbgp_maker.py

#!/usr/bin/env python3
""" Convert Github notebook URL to NBGitPuller URL

See: https://nbgitpuller.readthedocs.io/en/latest/link.html
"""

from argparse import ArgumentParser, RawDescriptionHelpFormatter
from warnings import warn
from urllib.parse import urlparse


DEFAULT_HUB = 'https://ds.lis.2i2c.cloud'


def to_url(url, hub_url, interface_prefix='tree', branch=None):
    prefix = f'{hub_url}/hub/user-redirect/git-pull?repo='
    up = urlparse(url)
    path_parts = up.path.split('/')[1:]
    gh_user, gh_repo, *other_parts = path_parts
    repo_path = f'{prefix}{up.scheme}%3A//{up.netloc}/{gh_user}/{gh_repo}'
    if not (other_parts):
        return repo_path + ('' if branch is None else f'&branch={branch}')
    if not other_parts[0] == 'blob':
        raise RuntimeError(
            'Expecting "blob" as third path component; try copying URL direct '
            'from Github interface')
    _, input_branch, *nb_parts = other_parts
    if branch and branch != input_branch:
        warn(f'URL branch "{input_branch}" does not match input argument '
             f'branch "{branch}" - preferring "{branch}".')
    branch = branch if branch else input_branch
    urlpath = '/'.join([interface_prefix, gh_repo] + nb_parts)
    return f'{repo_path}&urlpath={urlpath}&branch={branch}'


def get_parser():
    parser = ArgumentParser(description=__doc__,  # Usage from docstring
                            formatter_class=RawDescriptionHelpFormatter)
    parser.add_argument('url',
                        help='Github URL to repository or notebook')
    parser.add_argument('-u', '--hub-url', default=DEFAULT_HUB,
                        help=f'URL for base hub (default "{DEFAULT_HUB}"')
    parser.add_argument('-c', '--classic', action='store_true',
                        help='If set, use classic interface, not lab')
    parser.add_argument('-b', '--branch',
                        help=f'Branch (use Github default if not specified)')
    return parser


def main():
    parser = get_parser()
    args = parser.parse_args()
    interface_prefix = 'tree' if args.classic else 'lab/tree'
    print(to_url(args.url, args.hub_url, interface_prefix, args.branch))


if __name__ == '__main__':
    main()