github_backup.py

import boto3
from git import Repo
import os
import sys
from docopt import docopt
import shutil
import tarfile

GIT_API_KEY = os.getenv("GITHUB_API")
BUCKET = 'glomex-github-repo-backup'
DELIMITER = "=" * 50

def main():
    """
    Usage:
      github_backup.py single-repo --repo-name=<repo>
      github_backup.py multiple-repos --repo-file-list=<repo_file>
      github_backup.py (-h | --help)
      github_backup.py --version
    Commands:
       single-repo        Clone the repo and upload it into S3
       multiple-repos     Clone each repo from a file and upload it into S3
    Options:
      -h --help             Show this screen.
      --repo-name=<repo>            Name of repo that you want to backup
      --repo-file-list=<repo_file>  File with list of repos that you want to backup

    """
    args = docopt(main.__doc__, version="1.0")

    if args['single-repo']:
        repo_name = args['--repo-name']

        check_if_repo_exist(repo_name)
        clone_and_write_s3(repo_name)


    if args['multiple-repos']:
        repo_file_list = args['--repo-file-list']

        with open(repo_file_list, 'r') as file:
            repos_list = file.read().splitlines()

        for repo in repos_list:
            check_if_repo_exist(repo)
            clone_and_write_s3(repo)


def clone_and_write_s3(repo_name):
    session = boto3.Session()
    s3_client = session.resource('s3')

    repo_name_arch = repo_name + '.tar.gz'
    remote_url = "https://{}:[email protected]/glomex/{}".format(GIT_API_KEY, repo_name)

    print(DELIMITER + repo_name + DELIMITER)
    print("Cloning {} repo into {}".format(repo_name, os.getcwd()))
    Repo.clone_from(remote_url, repo_name)

    with tarfile.open(repo_name_arch, 'w:gz') as tar:
        tar.add(os.path.basename(repo_name), arcname=repo_name)

    try:
        print("Uploading {} file into {} bucket".format(repo_name + '.tar.gz', BUCKET))
        s3_client.Bucket(BUCKET).upload_file(repo_name_arch, repo_name_arch)
        clear_data(repo_name)
    except Exception as q:
        print('Cannot upload files %s' % q)
        print("Clearing uploaded data")
        clear_data(repo_name)
        sys.exit(1)

    print("Please go to the https://github.com/glomex/{}/settings and delete the repo".format(repo_name))


def clear_data(repo_name):
    print("Deleting cloned repo: {}".format(repo_name))
    shutil.rmtree(repo_name, ignore_errors=True)
    print("Deleted *.tar.gz file: {}".format(repo_name + '.tar.gz'))
    os.remove(repo_name + '.tar.gz')


def check_if_repo_exist(repo_name):
    if os.path.exists(repo_name):
        shutil.rmtree(repo_name)

if __name__ == "__main__":
    main()