devstats

#!/bin/bash
set -euo pipefail

SKIP_FETCH="${SKIP_FETCH:-}"
DEBUG="${DEBUG:-}"
NOSSH="${NOSSH:-}"

SCRIPTFOLDER="$(dirname "$(readlink -f "$0")")"

FROM="2018-01-01"
TO=$(date "+%Y-%m-%d")

if [ "${1-}" = "-h" ] || [ "${1-}" = "--help" ]; then
  echo "Usage $0 [YEAR-MONTH-DAY] [YEAR-MONTH-DAY]"
  echo "This script will fetch all Flatcar-related repos to ${SCRIPTFOLDER/..} and"
  echo "and show some dev stats."
  echo "The dates consist of year, month, and day (e.g., 2021-01-01)."
  echo "With only the start date given the time spans up to today."
  echo "Without the start date the time starts from ${FROM}."
  echo "Set NOSSH=1 as env var to clone the repos via https."
  echo "Set SKIP_FETCH=1 as env var to skip the repo update."
  echo "Set DEBUG=1 as env var to work on test data."
  exit 1
fi

if [ "${1-}" != "" ]; then
  FROM="${1}"
fi
if [ "${2-}" != "" ]; then
  TO="${2}"
fi

# Excluded: flatcar/ignition flatcar/afterburn flatcar/ign-converter flatcar/grub
# (because they are upstream projects)
REPOS=(flatcar/scripts flatcar/coreos-overlay flatcar/portage-stable flatcar/flatcar-website
       flatcar/init flatcar/flatcar-linux-update-operator flatcar/flatcar-build-scripts
       flatcar/mantle flatcar/update-ssh-keys flatcar/container-linux-config-transpiler
       flatcar/locksmith flatcar/bootengine flatcar/baselayout
       flatcar/flatcar-packer-qemu flatcar/sysext-bakery
       flatcar/flatcar-dev-util flatcar/torcx flatcar/update_engine flatcar/flatcar-terraform
       flatcar/coreos-cloudinit flatcar/fleetlock flatcar/toolbox flatcar/seismograph
       flatcar/updateservicectl flatcar/mayday
       flatcar/nss-altfiles flatcar/sysroot-wrappers flatcar/shim
       flatcar/fero flatcar/efunctions flatcar/chromite
       flatcar/flatcar-release-mirror flatcar/sdnotify-proxy
       flatcar/flog flatcar/ue-rs kinvolk/nebraska)

OUTPUT=""

if [ "${DEBUG}" = "1" ]; then
  OUTPUT="4 Some Body <somebody@microsoft.com>
3 Other Person <op@external.com>
7  Some Body <sb@users.noreply.github.com>
2  Another One <ao@kinvolk.io>"
fi

declare -A COUNTER
declare -A EMAILS
declare -A TO_REPOS

for REPO in "${REPOS[@]}"; do
  NAME=$(echo "${REPO}" | cut -d / -f 2)
  if [ "${DEBUG}" != "1" ]; then
    FOLDER="${SCRIPTFOLDER}/../${NAME}"
    if [ ! -d "${FOLDER}" ]; then
      echo "Cloning ${REPO} in ${FOLDER}"
      cd "${SCRIPTFOLDER}/.."
      URL="git@github.com:${REPO}.git"
      if [ "${NOSSH}" = "1" ]; then
        URL="https://github.com/${REPO}.git"
      fi
      git clone --recurse-submodules "${URL}" > /dev/null 2> /dev/null || { echo "Cloning failed, try again with NOSSH=1" >&2 ; exit 1 ;}
      cd -
    else
      if [ "${SKIP_FETCH}" != "1" ]; then
        echo "Fetching ${REPO} in ${FOLDER}"
        git -C "${FOLDER}" fetch origin > /dev/null 2> /dev/null
      fi
    fi
    REF="origin/main"
    if ! git -C "${FOLDER}" show "${REF}" > /dev/null 2> /dev/null; then
      REF="origin/flatcar-master"
    fi
    if ! git -C "${FOLDER}" show "${REF}" > /dev/null 2> /dev/null; then
      REF="origin/master"
    fi
    if ! git -C "${FOLDER}" show "${REF}" > /dev/null 2> /dev/null; then
      REF="origin/trunk"
    fi
    # Trim left space, use space not tabs, and ignore bots
    OUTPUT=$(git -C "${FOLDER}" shortlog -sne --since="${FROM}" --until="${TO}" "${REF}" | sed 's/^[ ]*//g' | tr '\t' ' ' | { grep -v -P '(flatcar-ci|Buildbot|dependabot|jenkins@localhost)' || true ; })
  fi
  while IFS= read -r line; do
    if [ "${line}" != "" ]; then
      # Format is "COUNT   NAME SURNAME <EMAIL>"
      COUNT=$(echo "${line}" | cut -d " " -f 1)
      AUTHOR=$(echo "${line}" | cut -d " " -f 2- | cut -d "<" -f 1)
      # Trim left and right space
      AUTHOR=$(echo "${AUTHOR}" | sed 's/^[ ]*//g' | sed 's/[ ]*$//g')
      EMAIL=$(echo "${line}" | cut -d " " -f 2- | cut -d "<" -f 2 | cut -d ">" -f 1)
      # Workarounds for a few names
      if [ "${AUTHOR}" = "yolossn" ] || [ "${AUTHOR}" = "Santhosh Nagaraj S" ]; then
        AUTHOR="S Santhosh Nagaraj"
      fi
      if [ "${AUTHOR}" = "Kai Lüke" ]; then
        AUTHOR="Kai Lueke"
      fi
      if [ "${AUTHOR}" = "william light" ]; then
        AUTHOR="William Light"
      fi
      if [ "${AUTHOR}" = "ashu8912" ]; then
        AUTHOR="Ashu Ghildiyal"
      fi
      if [ "${AUTHOR}" = "Iago López Galeiras" ]; then
        AUTHOR="Iago Lopez Galeiras"
      fi
      if [ "${AUTHOR}" = "krnowak" ]; then
        AUTHOR="Krzesimir Nowak"
      fi
      PREV_COUNT="${COUNTER["${AUTHOR}"]-0}"
      COUNTER["${AUTHOR}"]=$((COUNT + PREV_COUNT))
      PREV_EMAILS="${EMAILS["${AUTHOR}"]-}"
      if ! echo "${PREV_EMAILS}" | grep -q "${EMAIL}"; then
        EMAILS["${AUTHOR}"]=$(echo "${PREV_EMAILS} ${EMAIL}" | sed 's/^[ ]*//g')
      fi
      PREV_TO_REPOS="${TO_REPOS["${AUTHOR}"]-}"
      if ! echo "${PREV_TO_REPOS}" | grep -q "${NAME}"; then
        TO_REPOS["${AUTHOR}"]=$(echo "${PREV_TO_REPOS} ${NAME}" | sed 's/^[ ]*//g')
      fi
    fi
  done <<< "${OUTPUT}"
done

STATS=$(for AUTHOR in "${!COUNTER[@]}"; do
  echo "${COUNTER["${AUTHOR}"]} ${AUTHOR} <${EMAILS["${AUTHOR}"]-}> (${TO_REPOS["${AUTHOR}"]})"
done | sort -n -r)

PATTERN="(microsoft|kinvolk)"

echo
echo "Commits from ${FROM} to ${TO} in"
echo "  ${REPOS[*]}" | sed 's#flatcar/##g' | fmt
echo
echo "Kinvolk/Microsoft:"
echo "${STATS}" | grep -P "${PATTERN}"
echo
echo "Non-Kinvolk/Microsoft:"
echo "${STATS}" | grep -v -P "${PATTERN}"
echo
echo "Total commits: $(($(echo "${COUNTER[@]}" | tr -s '[:blank:]' '+')))"