Get repo URLs. Look one level deeper for git, hg or svn repositories, and get their original URLs

scan_repo_url.py

"""Look one level deeper for git, hg or svn repositories, and get their original URLs"""

import os
import subprocess
import argparse

def scan_repo():
    """Extract the URL of the repository"""
    get_git_url = 'git config --get remote.origin.url'
    get_hg_url = 'hg paths default'
    get_svn_url = "svn info | sed -ne 's/URL: //p'"
    try:
        url = subprocess.check_output(get_git_url, shell=True)
    except subprocess.CalledProcessError:
        try:
            url = subprocess.check_output(get_hg_url, shell=True)
        except subprocess.CalledProcessError:
            try:
                url = b'SVN:' + subprocess.check_output(get_svn_url, shell=True)
            except subprocess.CalledProcessError:
                print('Could not find git, hg or svn, on folder {}'.format(os.getcwd()))
                url = b''
    return url.decode(encoding='UTF-8')

def scan_one_level(root_dir, outfile):
    """For all directories in root_dir, get the URL,
    and save to outfile. (Not recursive).
    """
    cwd = os.getcwd()
    all_urls = []
    for cur_dir in os.listdir(root_dir):
        if os.path.isdir(root_dir+cur_dir):
            print('scanning {}'.format(root_dir+cur_dir))
            os.chdir(root_dir+cur_dir)
            all_urls.append(scan_repo())
    os.chdir(cwd)

    with open(outfile, 'w') as f:
        f.writelines(all_urls)


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('root_dir', help='Directory where you scan')
    parser.add_argument('outfile', help='File where URLs are saved')
    args = parser.parse_args()
    scan_one_level(args.root_dir, args.outfile)