#!/bin/sh

git log --since="3 years ago" \
        --numstat \
        --pretty=format:"COMMIT" |
awk '
/^COMMIT/ { next }

/^[0-9]/ {
    file = $3

    # Keep only hand-written core code
    if (file !~ /^(sw|sd|sc|vcl|desktop|framework)\//) next

    # Exclude generated / mechanical code
    if (file ~ /^include\/com\/sun\/star\//) next
    if (file ~ /^offapi\//) next
    if (file ~ /^udkapi\//) next
    if (file ~ /^workdir\//) next
    if (file ~ /^translations\//) next
    if (file ~ /^helpcontent2/) next
    if (file ~ /\.mk$/) next
    if (file ~ /^configure(\.ac|\.in)?$/) next
    if (file ~ /^download\.lst$/) next
    if (file ~ /^solenv\//) next

    commits[file]++
    churn[file] += $1 + $2
}
END {
    for (f in commits) {
        # Sweet spot: rarely touched, but not trivial
        if (commits[f] >= 3 && commits[f] <= 40 && churn[f] >= 20)
            printf "%4d commits | %6d churn | %s\n",
                   commits[f], churn[f], f
    }
}' |
sort -n