Quick script that generates some pretty stats about a git repository on a quarter-by-quarter basis. Written for some basic analysis of our many repositories over at https://rightmesh.io

gitqstats.sh

#!/usr/bin/env bash

# exit codes
EX_OK=0
EX_ERR=1
EX_USAGE=64

# default argument values
since_year=2015
until_year=$(date +%Y)
working_directory=$(pwd)
output_directory="$working_directory/git-stats"
node_env_version="7.10.1"
node_env_directory="/tmp/node7"
do_teardown=false

# usage() shows usage information
usage() {
    echo "Usage $0 [-s YEAR] [-u YEAR] [-C path] [-h]" 1>&2
    echo "Counts git repository statistics by quarter, outputting JSON data"
    echo ""
    echo "OPTIONS"
    echo "-s        the year to start counting stats by quarter (default: 1970)"
    echo "-u        the year to stop counting stats (default: current year)"
    echo "-C        specify the directory to change into (i.e. the repo)"
}

print_err() {
    echo "ERR: $1" 1>&2
}

# setup_env() sets up a Node v7 environment in /tmp/node7 to run the
# stats-gathering tools
setup_env() {
    which nodeenv 2>&1 1>/dev/null
    if [ $? -ne $EX_OK ]; then
        read -p "nodeenv is not installed, install it? [Y/n]" yn
        case $yn in
            n|N)
                echo "run pip install nodeenv to continue"
                return $EX_ERR
                ;;
            *)
                echo "pip install nodeenv..."
                pip install nodeenv
                if [ $? -ne $EX_OK ]; then
                    print_err "could not install nodeenv."
                    print_err "run pip install nodeenv to continue."
                    return $EX_ERR
                fi
        esac
    fi

    if [ ! -f "$node_env_directory/bin/activate" ]; then
        echo "Setting up Node7 environment..."
        nodeenv --node=$node_env_version "$node_env_directory" 1>/dev/null
        if [ $? -ne $EX_OK ]; then
            return $EX_ERR
        fi
    fi
    return $EX_OK
}

# get_emails() gets the list of all contributor emails in the repository,
# delimited by commas
get_emails() {
    if [ ! -d ".git" ]; then
        print_err "working directory is not a git repository"
        return $EX_ERR
    fi
    echo "$(git shortlog -sne | cut -d\< -f2 | sed 's/>//g' | tr '\n' ',')"
    return $?
}

# get_commits() gets the list of all commits for the given date range in the
# repository
get_commits() {
    echo "$(git log --all --abbrev-commit --pretty=oneline \
                --since="$1" --until="$2")"
    return $?
}

# main() is the main body of this script
main() {
    # parse arguments
    while getopts "s:u:C:hc" opt; do
        case "$opt" in
        s)
            since_year=${OPTARG:-$since_year}
            ;;
        u)
            until_year=${OPTARG:-$until_year}
            ;;
        C)
            working_directory=${OPTARG:-$working_directory}
            ;;
        c)
            do_teardown=true
            ;;
        h)
            usage
            exit $EX_OK
            ;;
        *)
            usage
            exit $EX_USAGE
            ;;
        esac
    done
    shift $((OPTIND - 1))

    if [ $working_directory != $(pwd) ]; then
        cd $working_directory
    fi

    # set up the environment, then source it
    if [ ! -d "$output_directory" ]; then
        mkdir -p "$output_directory"
    fi
    setup_env || exit $EX_ERR

    source "$node_env_directory/bin/activate"

    # check that git-stats, git-stats-importer are installed
    which git-stats 2>&1 1>/dev/null
    if [ $? -ne $EX_OK ]; then
        print_err 'git-stats is not installed'
        print_err 'run npm i -g git-stats to install it'
        exit $EX_ERR
    fi
    which git-stats-importer 2>&1 1>/dev/null
    if [ $? -ne $EX_OK ]; then
        print_err 'git-stats-importer is not installed'
        print_err 'run npm i -g git-stats-importer'
        exit $EX_ERR
    fi

    # import git stats
    emails=`get_emails`
    if [ $? -ne $EX_OK ]; then
        exit $EX_ERR
    fi
    git-stats-importer -e "$emails" 2>&1 1>/dev/null

    # get stats by quarter
    orig_since_year=$since_year
    while [ $since_year -le $until_year ]; do
        quarter=1
        while [ $quarter -le 4 ]; do
            out="$output_directory/$since_year-Q$quarter.html"
            date1=""
            date2=""
            case $quarter in
                1)
                    date1="$since_year Jan 1"
                    date2="$since_year Mar 31"
                    ;;
                2)
                    date1="$since_year Apr 1"
                    date2="$since_year June 30"
                    ;;
                3)
                    date1="$since_year Jul 1"
                    date2="$since_year Sept 30"
                    ;;
                4)
                    date1="$since_year Oct 1"
                    date2="$since_year Dec 31"
                    ;;
            esac
            # produce the stats graphs
            git-stats -g -s "$date1" -u "$date2" --raw \
                | tee "${out%.html}.json" \
                | git-stats-html -o "$out"

            sed -i 's/in the last year/in this quarter/g' $out

            # create a list of commits
            get_commits "$date1" "$date2" > "${out%.html}_commits.txt"

            quarter=$((quarter + 1))
        done
        since_year=$((since_year +1))
    done

    # create a histogram with gnuplot

    # tear down the environment and exit
    deactivate_node
    if [ "$do_teardown" == "true" ]; then
        rm -r "$node_env_directory"
    fi

    echo "Stats by quarter for years $orig_since_year-$until_year are available in "\
         "$output_directory."
    echo "You can view the HTML files in your browser."
}

main $@ || exit $EX_ERR