git-summary

#!/usr/bin/env bash


cd "$(git root)" || { echo "Can't cd to top level directory";exit 1; }

SUMMARY_BY_LINE=
DEDUP_BY_EMAIL=
for arg in "$@"; do
    case "$arg" in
        --line)
            SUMMARY_BY_LINE=1
            ;;
        --dedup-by-email)
            DEDUP_BY_EMAIL=1
            ;;
        -*)
            >&2 echo "unknown argument $arg found"
            exit 1
            ;;
        *)
            # set the argument back
            set -- "$@" "$arg"
            ;;
    esac

    shift
done

if [ -n "$DEDUP_BY_EMAIL" ] && [ -n "$SUMMARY_BY_LINE" ]; then
    >&2 echo "--dedup-by-email used with --line is not supported"
    exit 1
fi

commit=""
test $# -ne 0 && commit=$*
project=${PWD##*/}

#
# get date for the given <commit>
#

date() {
  # the $1 can be empty
  # shellcheck disable=SC2086
  git log --pretty='format: %ai' $1 | cut -d ' ' -f 2
}

#
# get active days for the given <commit>
#

active_days() {
  # shellcheck disable=SC2086
  date $1 | sort -r | uniq | awk '
    { sum += 1 }
    END { print sum }
  '
}

#
# get the commit total
#

commit_count() {
  # shellcheck disable=SC2086
  git log --oneline $commit | wc -l | tr -d ' '
}

#
# total file count
#

file_count() {
  git ls-files | wc -l | tr -d ' '
}

#
# remove duplicate authors who belong to the same email address
#

dedup_by_email() {
    # in:
    # 27  luo zexuan <luozexuan@xxx.com>
    #  7  罗泽轩 <luozexuan@xxx.com>
    # out:
    # 34 luo zexuan
    LC_ALL=C awk '
    {
        sum += $1
        if ($NF in emails) {
            emails[$NF] += $1
        } else {
            email = $NF
            emails[email] = $1
            # set commits/email to empty
            $1=$NF=""
            sub(/^[[:space:]]+/, "", $0)
            sub(/[[:space:]]+$/, "", $0)
            name = $0
            if (name in names) {
                # when the same name is associated with existed email,
                # merge the previous email into the later one.
                emails[email] += emails[names[name]]
                emails[names[name]] = 0
            }
            names[name] = email
        }
    }
    END {
        for (name in names) {
            email = names[name]
            printf "%6d\t%s\n", emails[email], name
        }
    }' | sort -rn -k 1
}

#
# list authors
#

format_authors() {
  # a rare unicode character is used as separator to avoid conflicting with
  # author name. However, Linux column utility will escape tab if separator
  # specified, so we do unesaping after it.
  LC_ALL=C awk '
  { args[NR] = $0; sum += $0 }
  END {
    for (i = 1; i <= NR; ++i) {
      printf "%s♪%2.1f%%\n", args[i], 100 * args[i] / sum
    }
  }
  ' | column -t -s♪ | sed "s/\\\x09/\t/g"
}

#
# fetch repository age from oldest commit
#

repository_age() {
  git log --reverse --pretty=oneline --format="%ar" | head -n 1 | LC_ALL=C sed 's/ago//'
}

#
# list the last modified author for each line
#
single_file() {
  while read -r data
  do
    if [[ $(file "$data") = *text* ]]; then
      git blame --line-porcelain "$data" 2>/dev/null | grep "^author\ " | LC_ALL=C sed -n 's/^author //p';
    fi
  done
}

#
# list the author for all file
#
lines() {
  git ls-files | single_file
}

#
# get the number of the lines
#
line_count() {
  lines | wc -l
}

# summary

echo
echo " project  : $project"

if [ -n "$SUMMARY_BY_LINE" ]; then
  echo " lines    : $(line_count)"
  echo " authors  :"
  lines | sort | uniq -c | sort -rn | format_authors
else

  # shellcheck disable=SC2046
  echo " repo age :" $(repository_age)
  # shellcheck disable=SC2086
  echo " active   :"  "$(active_days $commit)" days
  echo " commits  :" "$(commit_count)"
  if test "$commit" = ""; then
    echo " files    :" "$(file_count)"
  fi
  echo " authors  : "
  if [ -n "$DEDUP_BY_EMAIL" ]; then
    # the $commit can be empty
    # shellcheck disable=SC2086
    git shortlog -n -s -e $commit | dedup_by_email | format_authors
  else
    # shellcheck disable=SC2086
    git shortlog -n -s $commit | format_authors
  fi
fi