Print the top 10 Github contributors in some repos in an org since a given date

org_contributions.py

#!/usr/bin/env python3
# Note: For my simple use case, repo count is hardcoded to 3 repos from the same owner. Adjust this to whatever you need.

import sys
import time
from collections import defaultdict
from ghapi.all import GhApi

if len(sys.argv) < 7 or len(sys.argv) > 8:
    print("Usage: {} token owner repo1 repo2 repo3 start_date [verbose]".format(sys.argv[0]))
    sys.exit(1)

TOKEN = sys.argv[1]
OWNER = sys.argv[2]
repos = [(OWNER, repo) for repo in sys.argv[3:6]]
START_DATE = sys.argv[6]
verbose = len(sys.argv) == 8 and sys.argv[7] == "verbose"
api = GhApi(token=TOKEN)

def merge_dicts(dict1, dict2):
    merged_dict = dict1.copy()
    for key, value in dict2.items():
        if key in merged_dict:
            merged_dict[key] += value
        else:
            merged_dict[key] = value
    return merged_dict

def print_contributions(contributions_dict, repo):
    print("Top 10 contributors from {} onwards in {}:".format(START_DATE, repo))
    print("Rank\tContributor\tContributions")
    sorted_contributions = sorted(contributions_dict.items(), key=lambda x: x[1], reverse=True)[:10]
    max_len_contributor = max(len(contributor) for contributor, _ in sorted_contributions)
    for i, (contributor, count) in enumerate(sorted_contributions, start=1):
        print(f"{i}\t{contributor.ljust(max_len_contributor)}\t{count}")
    print("\n")

total_contributions = defaultdict(int)

for owner, repo in repos:
    try:
        contributors = api.repos.list_contributors(owner=OWNER, repo=repo)
    except Exception as e:
        print("Error fetching contributors:", e)
        sys.exit(1)

    contributors_logins = [contributor['login'] for contributor in contributors]

    print("Gathering data...")
    print(f"Contributions for {owner}/{repo} from {START_DATE} onwards:")

    repo_contributions_by_login = defaultdict(int)
    for contributor in contributors_logins:
        if contributor.endswith("bot"):
            continue  # Skip contributors with "bot" suffix
        try:
            search_result = api.search.issues_and_pull_requests(
                q=f"repo:{owner}/{repo} is:merged author:{contributor} created:>{START_DATE}"
            )
            contributions = search_result['total_count']
        except Exception as e:
            print(f"Error fetching contributions for {contributor}:", e)
            contributions = 0
        
        if verbose:
            print(f"{contributor}: {contributions}")
        
        repo_contributions_by_login[contributor] = contributions
        time.sleep(10)  # dumb hack to avoid API rate limiting

    total_contributions = merge_dicts(total_contributions, repo_contributions_by_login)
    print_contributions(contributions_dict=repo_contributions_by_login, repo=repo)

print_contributions(contributions_dict=total_contributions, repo="All repos")