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

org-contributors.go

package main

//
//	org-contributors.go
//	Created by Lightmann
//	February 2023
//

import (
	"os"
	"fmt"
	"sort"
	"bytes"
	"regexp"
	"os/exec"
	"strings"
	"strconv"
	"path/filepath"
	"github.com/jessevdk/go-flags"
	"golang.org/x/text/language"
	"golang.org/x/text/message"
)

var opts struct {
	Ignore string `short:"i" long:"ignore" description:"Comma-separated list of repos to ignore in the organization"`
	Org string `short:"o" long:"org" description:"Name of GitHub organization to query" required:"true"`
}

func main() {
	flags.Parse(&opts)

	// https://stackoverflow.com/a/30329351
	// Not off to a strong start here, GoLang ....
	cmd := fmt.Sprintf("curl -s \"https://api.github.com/users/%s/repos?per_page=100\" | grep -o 'git@[^\"]*'", opts.Org)
	cmdE := exec.Command("bash", "-c", cmd)
	// https://github.com/golang/go/issues/38268#issuecomment-609562062
	// https://stackoverflow.com/a/75169345
	// This seems harder than it has to be ....
	var buf bytes.Buffer
	cmdE.Stdout = &buf
	// Using the 'unused identifier' "_" to capture undesired err output
	_ = cmdE.Start()
	if err := cmdE.Wait(); err != nil {
		fmt.Printf("Error: the org specified -- %s -- does not appear to exist. Please check the name for typos.\n", opts.Org)
		return
	}

	repos := strings.Split(buf.String(),"\n")
	dir, _ := filepath.Abs(filepath.Dir(os.Args[0]))
	tmp, _ := os.MkdirTemp("", "")

	os.Chdir(tmp)

	var authors = make(map[string]int)
	LOOP: for _, repo := range repos {
		// Sanity check
		if len(repo) <= 0 {
			continue
		} else {
			buf.Reset()
		}

		// Switch repo url to https from ssh
		url := strings.Replace(repo, "git@github.com:", "https://github.com/", 1)

		// Skip specified repos
		if opts.Ignore != "" {
			filters := []string { opts.Ignore }
			if strings.Contains(opts.Ignore, ",") {
				filters = strings.Split(opts.Ignore, ",")
			}

			for _, filter := range filters {
				if strings.Contains(repo, filter) {
					fmt.Printf("[=] Skipping %s\n", url)
					continue LOOP
				}
			}
		}

		fmt.Printf("[x] Starting work on %s....\n", url)

		// Clone repo
		cmd = fmt.Sprintf("git clone --quiet --no-checkout %s", url)
		cmdE = exec.Command("bash", "-c", cmd)
		_ = cmdE.Start()
		_ = cmdE.Wait()

		// Grab repo name (i.e., directory name)
		bits := strings.Split(url, "/")
		directory := strings.TrimSuffix(bits[len(bits)-1], ".git")

		// Get contribution info
		cmd = fmt.Sprintf("git -C %s --no-pager log --stat", directory)
		cmdE = exec.Command("bash", "-c", cmd)
		cmdE.Stdout = &buf
		_ = cmdE.Start()
		_ = cmdE.Wait()
		commits := strings.Split(buf.String(), "commit ");

		fmt.Println("[+] Got the commit history. Quantifying diff....")
		for _, commit := range commits {
			lines := strings.Split(commit, "\n");

			// Grab relevant bits
			author := ""
			diff := ""
			for _, line := range lines {
				if strings.Contains(line, "Author: ") {
					author = line
				} else if strings.Contains(line, "changed, ") {
					diff = line
				}
			}

			// Santiy check
			if author == "" || diff == "" {
				continue
			}

			// Simplify bits
			author = strings.Replace(author, "Author: ", "", 1) // Remove prefix
			reg := regexp.MustCompile("<[^<>]*>")
			author = reg.ReplaceAllString(author, "${1}") // Remove email
			author = strings.TrimSpace(author) // Remove trailing space
			changes := strings.Split(diff, ", ")

			// Do some math
			for _, change := range changes {
				if strings.Contains(change, "-") || strings.Contains(change, "+") {
					reg = regexp.MustCompile("\\d")
					num, _ := strconv.Atoi(strings.Join(reg.FindAllString(change, -1)[:], ""))
					if _, exists := authors[author]; !exists {
						authors[author] = num
					} else {
						authors[author] += num
					}
				}
			}
		}
	}

	fmt.Println("[!] The results are in. Sorting....")
	// https://www.geeksforgeeks.org/how-to-sort-golang-map-by-keys-or-values/
	// Sort map keys according to respective val
	// Need to maintain the order, hence the new keys map
	keys := make([]string, 0, len(authors))
	for key := range authors {
		keys = append(keys, key)
	}
	sort.SliceStable(keys, func(i, j int) bool{
		return authors[keys[i]] > authors[keys[j]]
	})
	fmt.Printf("[-] Ordered list of contributors to %s repositories:\n", opts.Org)
	fmt.Println("-------------------------------------------------------")
	printer := message.NewPrinter(language.English)
	for _, author := range keys {
		// https://stackoverflow.com/a/46811454
		// Format each diff num nicely
		printer.Printf("◉ %s: %d edits\n", author, authors[author])
	}
	fmt.Println("-------------------------------------------------------")
	os.Chdir(dir)
	defer os.RemoveAll(tmp)
}