Get all open PRs in all repositories matching a regex.

get_open_prs.py

#!/usr/bin/python

from urllib2 import Request, urlopen, URLError
import sys
from argparse import ArgumentParser
import json
import re
from dateutil.parser import parse

MAX_RESULTS = 50

def get_all_repos(regex, token):
    page = 0
    matching_repos = []
    while True:
        url = "https://api.github.com/orgs/confluentinc/repos?type=private&page=%d&per_page=%d" % (page, MAX_RESULTS)
        request = Request(url)
        request.add_header("Accept", "application/vnd.github.symmetra-preview+json")
        request.add_header("Authorization", "token " + token)

        try:
            response = urlopen(request)
        except URLError as e:
            if hasattr(e, 'reason'):
                print("We failed to reach the API server. Reason: " + str(e.reason))
            elif hasattr(e, 'code'):
                print("The github server couldn't fulfill the request. Error code: " + str(e.code))
            sys.exit(1)

        pattern = re.compile(regex)

        repos = json.loads(response.read())
        if len(repos) < MAX_RESULTS:
            # We have reached the last page since we didn't fill the
            # quota
            break
        page = page + 1
        matching_repos.extend(filter(pattern.match, [repo.get('name') for repo in repos]))

    return matching_repos

def print_repo_pulls(repo, author_list, token):
    page = 0
    date_format = '%m/%d/%Y %H:%M:%S'
    while True:
        url = "https://api.github.com/repos/confluentinc/%s/pulls?page=%d&per_page=%d" % (repo, page, MAX_RESULTS)
        request = Request(url)
        request.add_header("Accept", "application/vnd.github.symmetra-preview+json")
        request.add_header("Authorization", "token " + token)

        try:
            response = urlopen(request)
        except URLError as e:
            if hasattr(e, 'reason'):
                print("We failed to reach the API server. Reason: " + str(e.reason))
            elif hasattr(e, 'code'):
                print("The github server couldn't fulfill the request. Error code: " + str(e.code))
            sys.exit(1)

        pulls = json.loads(response.read())
        if author_list and len(author_list) > 0:
            pulls = filter (lambda pull: pull['user']['login'] in author_list,
                            pulls)

        for pull_request in pulls:
            created = parse(pull_request['created_at']).strftime(date_format)
            updated = parse(pull_request['updated_at']).strftime(date_format)
            print "%s, %s, %s, %s, %s, %s" % (repo,
                                              created,
                                              updated,
                                              pull_request['user']['login'],
                                              pull_request['title'],
                                              pull_request['html_url'])


        if len(pulls) < MAX_RESULTS:
            # We have reached the last page since we didn't fill the
            # quota
            break
        page = page + 1


def main():

    parser = ArgumentParser()
    parser.add_argument("-r", "--repos", dest="repo_regex", required=True,
                        help="The name pattern for repos whose pull requests you want to list. eg. 'cc*'")
    parser.add_argument("-t", "--token", dest="github_token", required=True,
                        help="The github oauth token to use for authentication")
    parser.add_argument("-a", "--authors", dest="author_list", required=False, nargs='+',
                        help="Only return open PRs by these authors (github user names)")


    args = parser.parse_args()
    repos = get_all_repos(args.repo_regex, args.github_token)
    print "repo, created at, updated at, user, title, url"
    for repo in repos:
        print_repo_pulls(repo, args.author_list, args.github_token)


if __name__ == "__main__":
    main()