Clone all github repos

github-clone-by-user.py

# python github-clone-by-user.py username --skip-forks --excluded project1 project2
import requests
import subprocess
import argparse
import os


def clone_or_fetch_repo(username, token, skip_forks):
    headers = {'Authorization': f'token {token}'} if token else {}

    # First, clone or fetch the public repositories of the specified user
    base_url = f"https://api.github.com/users/{username}/repos"
    clone_or_fetch(base_url, headers, username, skip_forks)

    if not token:
        return

    # Then, clone or fetch the repositories that the authenticated user has access to
    base_url = "https://api.github.com/user/repos"
    clone_or_fetch(base_url, headers, username, skip_forks)

def clone_or_fetch(base_url, headers, username, skip_forks):
    page = 1
    while True:
        # Add the page and per_page parameters to the URL
        url = f"{base_url}?page={page}&per_page=100"

        # Send a GET request to the GitHub API
        response = requests.get(url, headers=headers)

        # If the request was successful
        if response.status_code == 200:
            # Get the JSON data from the response
            repos = response.json()

            # If there are no repositories in the response, we've reached the end of the list
            if not repos:
                break

            # For each repository
            for repo in repos:
                if skip_forks and repo.get('fork'):
                    print(f"Skipping forked repo {repo['name']}")
                    continue
                # Get the SSH URL
                clone_url = repo["ssh_url"]
                # Get the repository name
                repo_name = repo["name"]
                repo_owner = repo["owner"]["login"]

                if repo_owner != username:
                    continue

                if repo_name in excluded:
                    continue

                # Check if the repository already exists
                if os.path.exists(repo_name):
                    # If it does, fetch the updates
                    subprocess.run(["git", "-C", repo_name, "fetch"])
                else:
                    # If it doesn't, clone the repository
                    subprocess.run(["git", "clone", clone_url])

            # Go to the next page
            page += 1
        else:
            print(f"Failed to get repositories. Status code: {response.status_code}")
            break

# Create the parser
parser = argparse.ArgumentParser(description="Clone all repositories that the authenticated user has access to, from a specific GitHub user")

# Add the arguments
parser.add_argument('username', type=str, help="The GitHub username")
parser.add_argument('token', type=str, nargs='?', default=None, help="The GitHub personal access token (optional)")
parser.add_argument('--skip-forks', action='store_true', help="Whether to skip forked repositories (optional)")
parser.add_argument('--excluded', nargs='*', default=[], help="List of repository names to be excluded (optional)")

# Parse the arguments
args = parser.parse_args()
excluded = set(args.excluded)

# Call the function with the username and token arguments
clone_or_fetch_repo(args.username, args.token, args.skip_forks)