download github private repo readmes

ghreadme.py

# usage: python3 ghreadme.py
import json
import requests

GHUSERNAME = ''  # your username here
GHPASSWORD = ''  # your GH personal access token here
GHORG = 'department-of-veterans-affairs'  # change this for other orgs


class GHReadme(object):
    def __init__(self):
        self.ghusername = GHUSERNAME
        self.ghpassword = GHPASSWORD
        self.ghorg = GHORG
        self.api_base = 'https://api.github.com/'

    def _get_private_repos(self):
        """finds all private repos in org """
        private_repos = set()
        page = 1
        url = self.api_base + 'orgs/' + self.ghorg + '/repos?page=' + str(page) + '&per_page=100'
        while url:
            # import pdb; pdb.set_trace()
            print('Getting private repos - Page ' + str(page))
            r = requests.get(url, auth=(self.ghusername, self.ghpassword))
            json_content = json.loads(r.content)
            for i in range(len(json_content)):
                if json_content[i]['private']:
                    private_repos.add(json_content[i]['name'])
            try:
                url = r.links['next']['url']
            except KeyError:
                url = None
            page += 1
        print('\n' + str(len(private_repos)) + ' private repos found\n')
        return private_repos

    def _get_readme_links(self, repos):
        """resolves private repos to readme links"""
        repo_dict = {}
        for repo in repos:
            print('Getting README.md link for repo ' + repo)
            r = requests.get(self.api_base + 'repos/' + self.ghorg + '/' + repo + '/readme',
                             auth=(self.ghusername, self.ghpassword))
            json_repo = json.loads(r.content)
            try:
                repo_dict[repo] = json_repo['download_url']
            except KeyError:  # this is hacky
                pass
        print('\nGot links for ' + str(len(repo_dict)) + ' private repos\n')
        return repo_dict

    def _download_readmes(self, repos):
        """downloads all readmes to working diectory """
        count = 0
        for repo_name, readme_url in repos.items():
            print('Downloading README.md for ' + repo_name + ' as ' + repo_name + '_README.md')
            r = requests.get(readme_url, auth=(self.ghusername, self.ghpassword))
            with open(repo_name + '_README.md', 'wb') as f:
                f.write(r.content)
            count += 1
        print('\n' + str(count) + ' READMEs Downloaded')

    def run_task(self):
        """runs the loop to get private repos, get readme links, download """
        private_repo_list = self._get_private_repos()
        private_repo_dict = self._get_readme_links(private_repo_list)
        self._download_readmes(private_repo_dict)
        print('\nDONE!\n')


if __name__ == '__main__':
    g = GHReadme()
    g.run_task()