Unique active committer count for non-GitHub customers

unique-committers.sh

#! /usr/bin/env bash

# ***DISCLAIMER: Only tested on MacOS & Linux***
# Author: Dave Lacy (@davelacy)
# Created: 2022-05-17
# Updated: 2022-05-19

# Purpose: To provide a best guess estimate of the number of unique committers for GHAS pricing estimates. For organizations currently using 3rd party git version control products like ADO, GitLab, BitBucket, etc. 

# Before use, make sure you:
#   Have stashed your commits or committed your work
#   Clone the repos you want to include into one directory
#   Place this script in the root of the directory containing your cloned repos

# Options: 
#   --since YYYY-MM-DD (omitting this will result in a default of $today - 90 days) 
#   --branch master (omitting this the script will find the default branch for each repo)

# How to use:
#   1. cd into a directory containing the cloned repos you want to count active committers for
#   2. Add this script and execute it like `bash ./unique-commiters.sh` 
#   3. Should display an estimated count in the terminal output
#   4. Have a look at the output in the active_committers.txt file

# Examples:
#   bash ./unique-commiters.sh
#   bash ./unique-commiters.sh --since 2020-01-01
#   bash ./unique-commiters.sh --since 2020-01-01 --branch main

# Caveats:
#   If you have >1 committers with the same name (typically first and last), the count won't be accurate as there is a de-dupe method used

# create and overwrite active_committers.txt
echo "" > active_committers.txt

# get since and branch flags and set to variables
since=$(date -v -90d +%Y-%m-%d)

while [[ $# -gt 0 ]]
do
key="$1"

    case $key in
        --since)
        since="$2"
        shift # past argument
        ;;
        --branch)
        branch="$2"
        shift # past argument
        ;;
        *)
        # unknown option
        ;;
    esac
    shift # past argument or value
done

# print the date variable
echo "Finding active committers since: $since"

# if there are no directories in this directory, print error message
if [ "$(ls -d */)" = "" ]; then
    echo "Please ensure this script is run in a directory containing your repositories"
    exit 1
fi

git_command() {
    # if branch is not defined, use git command to find default branch
    if [ -z "$branch" ]; then
        git_branch=$(git rev-parse --abbrev-ref HEAD)
    else
        git_branch=$branch
    fi
    # branch = git remote show origin | sed -n '/HEAD branch/s/.*: //p'
    git checkout $git_branch && git pull && git log --pretty="%an" --since="$1" | sort | uniq >> ../active_committers.txt 
    echo "" >> ../active_committers.txt
}

# loop over each directory and run the git_command function from inside each directory
for directory in ./*/ ; do
    cd "$directory"
    git_command "$since"
    cd ..
done

# Remove the lines that include the word "dependabot"
awk '!/dependabot/' active_committers.txt > tmpfile && mv tmpfile active_committers.txt

# Remove duplicates
sort active_committers.txt | uniq -u > tmpfile && mv tmpfile active_committers.txt

# print the number of lines in the file active_committers.txt with a message 'Number of active committers: '
echo "Approximate number of active committers: $(wc -l active_committers.txt | awk '{print $1}')"