Git stats per dev (LOC) with delta and comparison

git-stats.sh

#!/bin/bash
# --- Config ---
EXCLUDE_EXTENSIONS="sql|log|csv|tsv|json|png|jpg|jpeg|gif|bmp|svg|mp4|mp3|zip|gz|tar|pdf|xlsx?|docx?"
WEEKS=4
COMPARE=false

# --- Args ---
for arg in "$@"; do
    if [[ "$arg" == "--compare" ]]; then
        COMPARE=true
    elif [[ "$arg" =~ ^[0-9]+$ ]]; then
        WEEKS="$arg"
    fi
done

# --- BSD-Compatible Dates ---
NOW=$(date +"%Y-%m-%d")
START=$(date -v -"${WEEKS}"w +"%Y-%m-%d")
PREV_START=$(date -v -"$((${WEEKS} * 2))"w +"%Y-%m-%d")

echo "Analyzing commits from $START to $NOW"
if $COMPARE; then
    echo "Comparing with previous $WEEKS weeks: $PREV_START to $START"
fi
echo

# --- Get authors as Name <email> using newline separation ---
authors=$(git log --no-merges --since="$START" --until="$NOW" --format='%aN <%aE>' | sort | uniq)

# Temp files for storing stats
tmp_now=$(mktemp)
tmp_prev=$(mktemp)
trap 'rm -f "$tmp_now" "$tmp_prev"' EXIT

# --- Helpers ---
human_format_value() {
    local num=$1
    local val_str=""

    if (( num < 1000 )); then
        val_str=$(printf "%d" "$num")
    elif (( num < 10000 )); then # 1000 to 9999 -> 1.0k to 9.9k
        val_str=$(awk -v n="$num" 'BEGIN{printf "%.1fk", n/1000}')
    elif (( num < 1000000 )); then # 10000 to 999999 -> 10k to 999k
        val_str=$(awk -v n="$num" 'BEGIN{printf "%.0fk", n/1000}')
    elif (( num < 10000000 )); then # 1,000,000 to 9,999,999 -> 1.0M to 9.9M
        val_str=$(awk -v n="$num" 'BEGIN{printf "%.1fM", n/1000000}')
    else # 10,000,000+ -> 10M, 123M etc.
        val_str=$(awk -v n="$num" 'BEGIN{printf "%.0fM", n/1000000}')
    fi
    echo "$val_str" | sed 's/\.0k/k/g; s/\.0M/M/g' # Clean up .0k and .0M
}

format_delta() {
    local num=$1
    local sign_char
    local abs_val_to_format

    if [[ "$num" -lt 0 ]]; then
        sign_char="-"
        abs_val_to_format=$((num * -1))
    else
        sign_char="+"
        abs_val_to_format=$num
    fi
    printf "%s%s" "$sign_char" "$(human_format_value "$abs_val_to_format")"
}

sign_commit_delta() {
    local num=$1
    if [[ "$num" -ge 0 ]]; then
        echo "+$num"
    else
        echo "$num"
    fi
}

get_stats() {
    local since="$1"
    local until="$2"
    local author="$3"
    
    # Extract email from "Name <email>" format for more reliable matching
    local email=$(echo "$author" | sed -n 's/.*<\([^>]*\)>.*/\1/p')
    if [[ -n "$email" ]]; then
        git log --no-merges --since="$since" --until="$until" --author="$email" --pretty=tformat: --numstat \
            | grep -vE "\.($EXCLUDE_EXTENSIONS)$" \
            | awk '{a+=$1; r+=$2} END {print (a ? a : 0), (r ? r : 0)}'
    else
        echo "0 0"
    fi
}

get_commits() {
    local since="$1"
    local until="$2"
    local author="$3"

    if [[ -n "$author" ]]; then
        git log --no-merges --since="$since" --until="$until" --format='%aN <%aE>' | grep -Fxc "$author"
    else
        echo "0"
    fi
}

# --- Collect stats ---
while IFS= read -r author; do
    [[ -z "$author" ]] && continue
    
    stats_now=($(get_stats "$START" "$NOW" "$author"))
    commits_now=$(get_commits "$START" "$NOW" "$author")
    
    echo "$author|$commits_now|${stats_now[0]:-0}|${stats_now[1]:-0}" >> "$tmp_now"
    
    if $COMPARE; then
        stats_prev=($(get_stats "$PREV_START" "$START" "$author"))
        commits_prev=$(get_commits "$PREV_START" "$START" "$author")
        echo "$author|$commits_prev|${stats_prev[0]:-0}|${stats_prev[1]:-0}" >> "$tmp_prev"
    fi
done <<< "$authors"

# --- Output ---
if $COMPARE; then
    printf "%-30s   %-25s   %-25s   %-25s   %-25s\n" "Author" "Commits (Now, Prev, Δ)" "Added (Now, Prev, Δ)" "Removed (Now, Prev, Δ)" "Changed (Now, Prev, Δ)"
else
    printf "%-30s %10s %9s %9s %9s\n" "Author" "Commits" "Added" "Removed" "Changed"
fi

while IFS= read -r author; do
    [[ -z "$author" ]] && continue
    
    display=$(echo "$author" | sed -E 's/ <.*>//')
    now=$(grep "^$(printf '%s' "$author" | sed 's/[[\.*^$()+?{|]/\\&/g')|" "$tmp_now" || echo "$author|0|0|0")
    IFS='|' read -r _ c_now a_now r_now <<< "$now"
    
    if $COMPARE; then
        prev=$(grep "^$(printf '%s' "$author" | sed 's/[[\.*^$()+?{|]/\\&/g')|" "$tmp_prev" || echo "$author|0|0|0")
        IFS='|' read -r _ c_prev a_prev r_prev <<< "$prev"
        
        c_diff=$((c_now - c_prev))
        a_diff=$((a_now - a_prev))
        r_diff=$((r_now - r_prev))
        
        ch_now=$((a_now + r_now))
        ch_prev=$((a_prev + r_prev))
        ch_diff=$((ch_now - ch_prev))

        # Format values for display
        c_diff_s=$(sign_commit_delta "$c_diff")

        a_now_h=$(human_format_value "$a_now")
        a_prev_h=$(human_format_value "$a_prev")
        a_diff_h=$(format_delta "$a_diff")

        r_now_h=$(human_format_value "$r_now")
        r_prev_h=$(human_format_value "$r_prev")
        r_diff_h=$(format_delta "$r_diff")

        ch_now_h=$(human_format_value "$ch_now")
        ch_prev_h=$(human_format_value "$ch_prev")
        ch_diff_h=$(format_delta "$ch_diff")

        printf "%-30s   %7d %7d %7s   %7s %7s %7s   %7s %7s %7s   %7s %7s %7s\n" \
            "$display" \
            "$c_now" "$c_prev" "$c_diff_s" \
            "$a_now_h" "$a_prev_h" "$a_diff_h" \
            "$r_now_h" "$r_prev_h" "$r_diff_h" \
            "$ch_now_h" "$ch_prev_h" "$ch_diff_h"
    else
        ch_now=$((a_now + r_now))
        a_now_h=$(human_format_value "$a_now")
        r_now_h=$(human_format_value "$r_now")
        ch_now_h=$(human_format_value "$ch_now")
        printf "%-30s %10s %9s %9s %9s\n" "$display" "$c_now" "$a_now_h" "$r_now_h" "$ch_now_h"
    fi
done <<< "$authors"