git-wordiness

Updated version of the old git-wordiness script

#!/usr/bin/env bash

# FIXME: Names with non-ASCII characters can throw off alignment of
# columns.

usage() {
    cat <<HELPTXT 1>&2
git-wordiness [OPTIONS]

List general sums about commit messages (i.e., wc output).

OPTIONS
-s[PREFIX]
    Pass all names through sha1sum (with an optional PREFIX).
    Can be used to mask names, if that's necessary.
-csv
    Output results in comma-separated value format.
-tsv
    Output results in tab-separated value format.
-order=ORDER
    Set the output ordering to one of the following:
     - commits, c
       Order by total commits.
     - wpc, W
       Order by words-per-commit
     - lines, l
       Order by lines used in all commit messages
       (does not rewrap or unbreak paragraphs).
     - words, w
       Order by total words used in all commit messages.
     - bytes, b
       Order by bytes used for all commit messages.
     - name, n
       Order by committer name.
HELPTXT
}

set_order() {
        local oarg="$1"
        case "$oarg" in
        commits|c)
                 order=1
                 ;;
        wpc|W)
                order=2
                ;;
        lines|l)
                order=3
                ;;
        words|w)
                order=4
                ;;
        bytes|b)
                order=5
                ;;
        name|n)
                order=6
                ordertype='-d'
                ;;
        *)
                printf "unrecognized ordering '%s'\n" "$oarg" 1>&2
                exit 1
                ;;
        esac
}

hfmt='%10s %10s %10s %10s %10s %s\n'
fmt='%10d %10.2f %10d %10d %10d %s\n'
head=1
order=1
ordertype='-n'
shanames=0
shaprefix=''
sep=' '
logargs=(--all)
while [[ $# -gt 0 ]] ; do
        arg="$1"
        shift
        case "$arg" in
        -h|-help|--help)
                usage
                exit 0
                ;;
        -s=*)
                shanames=1
                shaprefix="${arg#-s=}"
                ;;
        -s*)
                shanames=1
                shaprefix="${arg#-s}"
                ;;
        -c|-csv|--csv)
                hfmt='%s,%s,%s,%s,%s,%s\n'
                fmt='%d,%.2f,%d,%d,%d,%s\n'
                sep=','
                ;;
        -t|-tsv|--tsv)
                hfmt='%s\t%s\t%s\t%s\t%s\t%s\n'
                fmt='%d\t%.2f\t%d\t%d\t%d\t%s\n'
                sep="$(printf '\t')"
                ;;
        -order=*) set_order "${arg#-order=*}" ;;
        --order=*) set_order "${arg#--order=*}" ;;
        -o?) set_order "${arg:2}" ;;
        -nh|-nohead|--nohead) head=0 ;;
        --)
                logargs=("$@")
                break
                ;;
        *)
                logargs=("$arg" "$@")
                break
                ;;
        esac
done

export shanames
export shaprefix

export logquote="$(if [[ "${#logargs[@]}" -gt 0 ]] ; then printf ' %q' "${logargs[@]}" ; fi)"

export fmt
if [[ $head == 1 ]]
then
        printf "$hfmt" COMMITS WPC LINES WORDS BYTES NAME
fi
git log --pretty='%aN' --use-mailmap "${logargs[@]}" |
        sort |
        uniq |
        xargs -I @@ \
                env NAME=@@ REGEX='^@@ <' \
                bash -c $'
                    wcout=(
                        # lines words bytes
                        $(git log $logquote --use-mailmap --pretty="%B" --author="$REGEX" | wc)
                    )
                    commits=$(git log $logquote --use-mailmap --pretty="%H" --author="$REGEX" | wc -l)
                    comp=(
                        "$commits"
                        "$(bc <<< "scale=2; ${wcout[1]}/${commits}")"
                        "${wcout[@]}"
                    )
                    if [ $shanames -eq 1 ]; then
                        NAME="$(printf %s%s "$shaprefix" "$NAME" | sha1sum | awk \'{print $1}\')"
                    fi
                    printf "$fmt" "${comp[@]}" "$NAME"
                ' | \
        sort -b $ordertype -k "${order}n"