Get a list of contributors to a public GitHub org sorted by total diff

org-contributors.py

#!/usr/bin/env python3
#
# 	org-contributors.py
#	Created by Lightmann
#	February 2023
#

import subprocess, tempfile, argparse, sys, os, re

def help_prompt():
    """Usage: org-contributors.py [options]
Options:
  [-i|--ignore]		Comma-separated list of repos to ignore in the organization
  [-o|--org] 		Name of GitHub organization to query
  [-h|--help]		Display this page"""

    print(help_prompt.__doc__)

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("-i", "--ignore", action="store")
    parser.add_argument("-o", "--org", action="store", required=True)
    parser.print_help = help_prompt
    opt = parser.parse_args()

    curl = subprocess.getoutput(f"curl -s \"https://api.github.com/users/{opt.org}/repos?per_page=100\" | grep -o 'git@[^\"]*'")
    if curl == "":
        print(f"Error: the org specified -- {opt.org} -- does not appear to exist. Please check the name for typos.")
        exit(1)

    repos = curl.split("\n")
    cdir = sys.path[0]
    tmp = tempfile.TemporaryDirectory()
    os.chdir(tmp.name)
    authors = {}
    for repo in repos:
        # Switch repo url to https from ssh
        url = repo.replace("git@github.com:", "https://github.com/")

        # Skip specified repos
        if opt.ignore:
            filters = [opt.ignore]
            if "," in opt.ignore:
                filters = opt.ignore.split(",")

            # https://stackoverflow.com/q/653509
            # Check if filter strings are present
            class skip(Exception): pass
            try:
                for i in filters:
                    if i in repo:
                        print(f"[=] Skipping {url}")
                        raise skip
            except skip:
                continue

        print(f"[x] Starting work on {url}....")

        # Clone repo
        os.system(f"git clone --quiet --no-checkout {url}")

        # Grab repo name (i.e., directory name)
        # Split url at "/", grab repo, and remove extension
        directory = (url.split("/")[-1]).replace(".git", "")

        # Get contribution info
        log = subprocess.getoutput(f"git -C {directory} --no-pager log --stat")
        commits = log.split("commit ")

        print("[+] Got the commit history. Quantifying diff....")
        for commit in commits:
            lines = commit.split("\n")

            # Grab relevant bits
            author = ""
            diff = ""
            for line in lines:
                if "Author: " in line:
                    author = line
                elif "changed, " in line:
                    diff = line

            # Santiy check
            if not author or not diff:
                continue

            # Simplify bits
            author = author.replace("Author: ", "") # Remove prefix
            author = re.sub("<[^<>]*>", "", author) # Remove email
            author = author.strip() # Remove trailing space
            changes = diff.split(", ")

            # Do some math
            for change in changes:
                if "-" in change or "+" in change:
                    num = int(re.sub("\D", "", change))
                    if num != "":
                        if not author in authors:
                            authors[author] = num
                        else:
                            authors[author] += num

    print("[!] The results are in. Sorting....")
    # https://stackoverflow.com/a/2258273
    # Sort dict key-val pairs according to vals
    # Puts into tuples in a list to maintain order
    sortedA = sorted(authors.items(), key=lambda x: x[1], reverse=True)
    print(f"[-] Ordered list of contributors to {opt.org} repositories:")
    print("-------------------------------------------------------")
    for pair in sortedA:
        # https://stackoverflow.com/a/25281291
        # Format each diff num nicely
        diff = (format(pair[1], ",d"))
        print(f"◉ {pair[0]}: {diff} edits")

    print("-------------------------------------------------------")
    os.chdir(cdir)
    tmp.cleanup()

if __name__ == "__main__":
    main()