Migrate from github to self-hosted gitlab

migration.py

import subprocess
import requests
import os


def call(cmd):
    print('Executing', cmd)
    subprocess.call(cmd, shell=True)


def cd(d):
    print('Changing dir', d)
    os.chdir(os.path.join(os.getcwd(), d))


with open('repos.txt') as f:
    # every line is a repo name
    content = f.readlines()
    content = [x.strip() for x in content]

    # save initial working dir
    cwd = os.getcwd()

    for repo in content:
        # always start from initial working dir
        os.chdir(cwd)

        # clone from github
        clone_cmd = 'git clone git@github.com:<YOUR_ORGANIZATION>/{}.git'.format(repo)
        call(clone_cmd)

        # get into the repo folder
        cd(repo)

        # remove gitlab remote, in case you're running this script twice
        remote_remove_cmd = 'git remote remove gitlab'
        call(remote_remove_cmd)

        # track all remote branches
        track_cmd = "git branch -r | grep -v '\->' | while read remote; do git branch --track \"${remote#origin/}\" \"$remote\"; done"  # noqa
        call(track_cmd)

        # pull all github branches
        fetch_cmd = 'git fetch --all'
        pull_cmd = 'git pull --all'
        call(fetch_cmd)
        call(pull_cmd)

        # create gitlab repo
        # Gather namespace_id from https://gitlab.com/api/v4/namespaces?search=<YOUR_ORGANIZATION>
        gitlab_endpoint = 'https://gitlab.com/api/v4/projects?name={}&namespace_id=<NAMESPACE_ID>'  # noqa
        gitlab_access_token = '<YOUR_GITLAB_ACCESS_TOKEN>'
        gitlab_response = requests.post(
            gitlab_endpoint.format(repo),
            headers={
                'PRIVATE-TOKEN': gitlab_access_token,
            })
        print(gitlab_response.json())

        # add gitlab remote
        remote_cmd = 'git remote add gitlab ssh://git@gitlab.com:22/<YOUR_ORGANIZATION>/{}.git'.format(repo)  # noqa
        call(remote_cmd)

        # push to gitlab
        push_cmd = 'git push --all gitlab'
        call(push_cmd)

repos.txt

repo1
repo2
repo3
repo4
repo5

It could be improved by:

• connecting to the Github API and gathering the repos instead of using a text file
• gathering the namespace_id from the Gitlab API