Print the top 10 Github contributors to a repo since a given date

contributions.py

#!/usr/bin/env python3

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

if len(sys.argv) < 5 or len(sys.argv) > 6:
    print("Usage: {} token owner repo start_date [verbose]".format(sys.argv[0]))
    sys.exit(1)

TOKEN = sys.argv[1]
OWNER = sys.argv[2]
REPO = sys.argv[3]
CREATED = sys.argv[4]
verbose = len(sys.argv) == 6 and sys.argv[5] == "verbose"

api = GhApi(token=TOKEN)

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...")

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:>{CREATED}"
        )
        contributions = search_result['total_count']
    except Exception as e:
        print(f"Error fetching contributions for {contributor}:", e)
        contributions = 0
    
    if verbose:
        print("{}: {}".format(contributor, contributions))
    
    contributions_by_login[contributor] = contributions
    time.sleep(5)  # dumb hack to avoid API rate limiting

print("Top 10 contributors from {} onwards:".format(CREATED))

sorted_contributions = sorted(contributions_by_login.items(), key=lambda x: x[1], reverse=True)[:10]

max_len_contributor = max(len(contributor) for contributor, _ in sorted_contributions)
print("Rank\tContributor\tContributions")
for i, (contributor, count) in enumerate(sorted_contributions, start=1):
    print(f"{i}\t{contributor.ljust(max_len_contributor)}\t{count}")