github-issues-archive.py

#!/usr/bin/env python3
#
# Downloads all issues with comments for a single GitHub repo.
#
# - Requires Python 3 and the Requests library.
# - Merges comments into issues and saves as a single JSON file.
# - Uses basic auth so it works with both personal and organsiation repos.
#

import json
import requests

repo = ''    # Name of the target repository 
org = ''     # Repo owner - user or org
user = ''    # User account login, same as org if for a personal account 
secret = ''  # User account Password

def main():
    global repo, org

    print('Listing issues...', end='', flush=True)
    issues = list_all(
        'https://api.github.com/repos/{}/{}/issues'.format(org,repo),
        state='all'
    )
    print(' found:', len(issues))

    for issue in issues:
        number = str(issue['number'])
        print('Listing comments for issue #{}...'.format(number), end='', flush=True)
        comments = list_all(issue['comments_url'])
        print(' found:', len(comments))
        issue['comments'] = comments

    filename = '{}_{}_github_issues.json'.format(org, repo)
    print('Writing to: ', filename)
    json.dump(issues, open(filename, 'w'), indent=4)

def list_all(url, **query_args):
    global user, secret
    query_args['per_page'] = '100'
    items = []
    page = 1
    while True:
        query_args['page'] = str(page)
        response = requests.get(url, params=query_args, auth=(user,secret))
        #print(response.url)
        new_items = response.json()
        if len(new_items) == 0:
            break
        items += new_items
        page += 1
    return items

if __name__ == '__main__':
    main()