Collects all pulls and related information from GitHub

get_github_pulls.py

import argparse
import json
from tqdm.auto import tqdm
from agithub.GitHub import GitHub


def get_pulls(client, owner, repository):
    pulls = []
    status, pulls = client.repos[owner][repository].pulls.get(state='all')
    if status != 200:
        raise Exception(status)
    return pulls


def get_commits(client, owner, repository, pulls):
    commits = dict()
    for pull in tqdm(pulls, desc='Commits'):
        pr_number = pull['number']
        status, pr_commits = client.repos[owner][repository].pulls[pr_number].commits.get(
        )
        if status != 200:
            raise Exception(status)
        commits[pr_number] = pr_commits
    return commits


def get_timelines(client, owner, repository, pulls):
    timelines = dict()
    for pull in tqdm(pulls, desc='timelines'):
        pr_number = pull['number']
        status, timeline = client.repos[owner][repository].issues[pr_number].timeline.get(
        )
        if status != 200:
            raise Exception(status)
        timelines[pr_number] = timeline
    return timelines


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='desc')
    parser.add_argument('owner', type=str)
    parser.add_argument('repository', type=str)
    parser.add_argument('token', type=str)
    parser.add_argument('--pulls', '-p', dest='pulls_file',
                        default='./pulls.json')
    parser.add_argument('--cached_pulls', '-cp',
                        dest='cached_pulls_file', default=None)
    parser.add_argument('--commits', '-c', dest='commits_file', default=None)
    parser.add_argument('--timeline', '-t', dest='timeline_file', default=None)
    args = parser.parse_args()

    client = GitHub(token=args.token, paginate=True)
    owner = args.owner
    repository = args.repository

    pulls = []

    if args.cached_pulls_file:
        print('Load pulls from cache file {}...'.format(args.cached_pulls_file))
        with open(args.cached_pulls_file, 'r') as json_file:
            pulls = json.load(json_file)
    else:
        print('Load pulls from {0}/{1}...'.format(owner, repository))
        pulls = get_pulls(client, owner, repository)
        with open(args.pulls_file, 'w') as json_file:
            json.dump(pulls, json_file)

    if args.commits_file:
        print('Load pulls from {0}/{1}...'.format(owner, repository))
        commits = get_commits(client, owner, repository, pulls)
        with open(args.commits_file, 'w') as json_file:
            json.dump(commits, json_file)

    if args.timeline_file:
        print('Load pulls from {0}/{1}...'.format(owner, repository))
        timelines = get_timelines(client, owner, repository, pulls)
        with open(args.timeline_file, 'w') as json_file:
            json.dump(timelines, json_file)