Clone a git repository and retain directory structure

git-dolly

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
#
# git-dolly -- Clone a git repository and retain directory structure
#
# Uses by default "~/src" as the root to where to clone repositories. It then
# automatically sets up the target directory based on the given URL.
#
# To override the default, set GIT_DOLLY_PATH.
#
# Example:
#
# These commands would all clone the repository to $HOME/src/scm.com/path/repository.
# $ git dolly https://scm.com/path/repository.git
# $ git dolly scm.com:path/repository.git

import os
import subprocess
import sys

from urllib.parse import urlparse


def main():
    root = os.getenv('GIT_DOLLY_PATH', os.path.expanduser('~/src'))
    prog = os.path.basename(sys.argv[0])
    argv = sys.argv[2:] if sys.argv[0] == 'git' else sys.argv[1:]

    repo_path = get_repo_path(root, argv[-1])
    dirname = os.path.dirname(repo_path)
    if os.path.exists(repo_path) and len(os.listdir(repo_path)) > 0:
        sys.stderr.write("{0}: repository exists: {1}\n".format(prog, repo_path))
        return 1

    if not os.path.exists(dirname):
        os.makedirs(dirname)
    return subprocess.call(['git', 'clone'] + argv + [repo_path])


def get_repo_path(root, repo):
    """Get repository filesystem path

    Determine the format of the git repository URL and figure out the hostname,
    path and name of repository to create the path where the repository should
    be stored.

    >>> get_repo_path('/usr/src', 'https://scm.com/x/y')
    '/usr/src/scm.com/x/y'
    >>> get_repo_path('/usr/src', 'https://[email protected]/x/y')
    '/usr/src/scm.com/x/y'
    >>> get_repo_path('/usr/src', 'ssh://[email protected]/x/y')
    '/usr/src/scm.com/x/y'
    >>> get_repo_path('/usr/src', 'ssh+x://[email protected]/x/y')
    '/usr/src/scm.com/x/y'
    >>> get_repo_path('/usr/src', 'git://scm.com/x/y')
    '/usr/src/scm.com/x/y'
    >>> get_repo_path('/usr/src', 'scm.com:x/y')
    '/usr/src/scm.com/x/y'
    >>> get_repo_path('/usr/src', '[email protected]:x/y')
    '/usr/src/scm.com/x/y'
    """
    p = urlparse(repo)
    if p.netloc:
        host, path = p.netloc, p.path
    else:
        host, path = repo.split(':', 1)
    host = host.split('@')[-1]
    assert host and ':' not in host, 'unsupported url'
    dirname = os.path.dirname(path)
    basename = os.path.basename(path).rstrip('.git')
    return os.path.join(root, host.lstrip('/'), dirname.lstrip('/'), basename)


if __name__ == '__main__':
    import doctest
    doctest.testmod()
    sys.exit(main())