Python script getting owners of repos with outside collaborators

github_org_outside_collaborators.py

import csv
import json
from urllib import request

usage = """
------------------------
Usage
------------------------
Add your Github Access Token with "read_user, user:email, public_repo, repo:invite" scopes like so;
token = "ksdjhkglkfjhløaksjbli378gblaup"

Usage;
$ python3 ./outside_collaborators.py

This will create a .csv file with every owner of a repository which has outside collaborators.
"""
github_admins = ("knut-erik", "ingvald", "pgdr", "solomson", "StigNil", "roar-statoil", "MerMyhr")
token = ""

if not token:
    print(f"ERROR: Access token not given.")
    print(usage)
    print("Exiting...")
    exit(1)


def collabs_query(end_cursor):
    graphql_query = f"""
    query{{organization(login: "equinor"){{
        repositories(first: 100{end_cursor}){{
          nodes{{
            name,
            description,
            url,
            collaborators(affiliation: OUTSIDE){{
              nodes{{
                login,
                company,
                name,
                email
              }}
            }}
    
          }},
          pageInfo{{
            endCursor,
            hasNextPage
          }}
        }}
    }}
    }}
    """
    post_data = json.dumps({"query": graphql_query}).encode()
    return post_data


def owners_query(repo):
    graphql_query = f"""
    {{
      organization(login: "equinor") {{
        repository(name: "{repo}") {{
          name
          description
          url
          collaborators(affiliation: ALL) {{
            edges {{
              node {{
                login
                email
              }}
              permission
            }}
            totalCount
          }}
        }}
      }}
    }}
    """
    post_data = json.dumps({"query": graphql_query}).encode()
    return post_data


def get_outside_colabs():
    print("Getting all outside collaborators in the Equinor organization")
    url = 'https://api.github.com/graphql'
    headers = {'Accept': 'application/vnd.github.v3+json',
               'Authorization': 'Bearer ' + token}
    end_cursor = ""
    outsiders = {}

    while True:

        req = request.Request(url=url, headers=headers, data=collabs_query(end_cursor))
        response = request.urlopen(req).read().decode()
        users_page = json.loads(response)

        cursor = users_page['data']['organization']['repositories']['pageInfo']['endCursor']
        end_cursor = ", after: \"" + cursor + "\""

        repositories = [r for r in users_page["data"]["organization"]["repositories"]["nodes"]]
        for repo in repositories:
            collaborators = repo.get('collaborators')
            users = collaborators.get("nodes") if collaborators else None
            if users:
                for user in users:
                    outsiders[user["login"]] = {
                        "repo": repo["name"],
                        "email": user["email"],
                        "login": user["login"],
                        "name": user["name"],
                        "company": user["company"]
                    }

        if not users_page['data']['organization']['repositories']['pageInfo']['hasNextPage']:
            break
    print(f"{len(outsiders)} total outside collaborators in the Equinor Org")
    return outsiders


def get_repo_owners(repo_list):
    print("Getting all owners of every repository with an outside collaborator")
    url = 'https://api.github.com/graphql'
    headers = {'Accept': 'application/vnd.github.v3+json',
               'Authorization': 'Bearer ' + token}
    repo_owners_w_outsiders = []

    for repo in repo_list:

        req = request.Request(url=url, headers=headers, data=owners_query(repo))
        response = request.urlopen(req).read().decode()
        users_page = json.loads(response)

        collaborators = [c for c in users_page["data"]["organization"]["repository"]["collaborators"]["edges"]]
        for user in collaborators:
            login = user["node"]["login"]
            role = user["permission"]
            email = user["node"]["email"]

            if login not in github_admins and role is "ADMIN":
                repo_owners_w_outsiders.append({"repo": repo, "role": role, "login": login, "email": email})

    return repo_owners_w_outsiders


if __name__ == "__main__":
    with open("collab_repo_owners.csv", "w") as file:
        fieldnames = ['repo', "role", 'email', 'login']
        writer = csv.DictWriter(file, fieldnames=fieldnames)
        writer.writeheader()
        outsiders = get_outside_colabs()
        repos_with_outsiders = []
        [repos_with_outsiders.append(o["repo"]) for o in outsiders.values() if o["repo"] not in repos_with_outsiders]

        repo_owners = get_repo_owners(repos_with_outsiders)

        for user in repo_owners:
            writer.writerow(user)

Notes

• Written in Python version 3.8, so you should probably use that as well.
• No dependencies! No need for pip install, venv etc.