#!/usr/bin/env bash

# Shows the largest object of a Git repository:
# - Size
# - Object ID
# - Path
# - Branch(es), the object
#
# Usage:
#   ./git-biggest-objects.sh [NUMBER]
#
# Example:
#   ./git-biggest-objects.sh 30

set -euo pipefail

COUNT="${1:-20}"

if ! git rev-parse --git-dir >/dev/null 2>&1; then
    echo "Fehler: Kein Git-Repository."
    exit 1
fi

TMP_OBJECTS="$(mktemp)"
TMP_BRANCHES="$(mktemp)"

cleanup() {
    rm -f "$TMP_OBJECTS" "$TMP_BRANCHES"
}
trap cleanup EXIT

echo "Collecting objects ..."
git rev-list --objects --all > "$TMP_OBJECTS"

echo "Collecting branches ..."
while read -r commit branches; do
    for branch in $branches; do
        echo "$commit $branch"
    done
done < <(
    git for-each-ref \
        --format='%(objectname) %(refname:short)' \
        refs/heads
) > "$TMP_BRANCHES"

echo
printf "%-12s %-40s %-50s %s\n" "SIZE" "OBJECT ID" "BRANCH" "PATH"
printf "%-12s %-40s %-50s %s\n" "----" "---------" "------" "-----"

git cat-file --batch-check='%(objectname) %(objecttype) %(objectsize)' < <(
    awk '{print $1}' "$TMP_OBJECTS"
) |
awk '$2 == "blob"' |
sort -k3 -n |
tail -n "$COUNT" |
while read -r object type size; do

    path="$(grep "^$object " "$TMP_OBJECTS" | sed "s/^$object //")"

    # Find commit(s) that contain(s) the object
    commits="$(git log --all --find-object="$object" --format='%H' 2>/dev/null || true)"

    branches=""

    if [[ -n "$commits" ]]; then
        while read -r commit; do
            found_branches="$(git branch --contains "$commit" --format='%(refname:short)' 2>/dev/null || true)"
            branches="${branches} ${found_branches}"
        done <<< "$commits"
    fi

    # Remove duplicates
    branches="$(echo "$branches" | tr ' ' '\n' | sort -u | tr '\n' ',' | sed 's/,$//')"

    # Human-readable size
    human_size="$(numfmt --to=iec-i --suffix=B "$size")"

    printf "%-12s %-40s %-50s %s\n" \
        "$human_size" \
        "$object" \
        "${branches:-?}" \
        "$path"

done | sort -h