#!/usr/bin/env bash
# manifest_gen.sh — minimal manifest generator using find + md5sum
# Default output: /tmp/manifest-<hostname>.txt
# Format: "<md5>  <absolute_path>"

set -euo pipefail

HOST="$(hostname -s 2>/dev/null || hostname 2>/dev/null || echo unknown)"
OUT="./manifest-${HOST}.txt"
ROOT="${1:-/}"

# Default excludes (edit here)
EXCLUDES=(/proc /sys /dev /run /tmp /var/run /var/lock /lost+found /mnt /media /cloudian* /var/lib/cassandra)

ERR="${OUT}.err"
: > "$OUT"
: > "$ERR"

ROOT="$(cd "$ROOT" && pwd -P)"

{
  echo "# manifest_version=md5sum-v1"
  echo "# created_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)"
  echo "# host=$HOST"
  echo "# root=$ROOT"
  echo -n "# excludes="
  printf "%s " "${EXCLUDES[@]}"
  echo
  echo "# format=<md5>  <absolute_path>"
} >> "$OUT"

# Build prune expression
PRUNE=()
for ex in "${EXCLUDES[@]}"; do
  # normalize excludes if they exist; if they don't, still keep as-is
  if [ -e "$ex" ]; then
    ex="$(cd "$ex" 2>/dev/null && pwd -P || echo "$ex")"
  fi
  PRUNE+=( -path "$ex" -o -path "$ex/*" -o )
done
unset 'PRUNE[${#PRUNE[@]}-1]'  # drop last -o

# Deterministic ordering if available (GNU sort -z)
if sort -z </dev/null >/dev/null 2>&1; then
  find "$ROOT" \( "${PRUNE[@]}" \) -prune -o -type f -print0 2>>"$ERR" \
    | sort -z 2>>"$ERR" \
    | xargs -0 -r md5sum -- 2>>"$ERR" >> "$OUT"
else
  find "$ROOT" \( "${PRUNE[@]}" \) -prune -o -type f -print0 2>>"$ERR" \
    | xargs -0 -r md5sum -- 2>>"$ERR" >> "$OUT"
fi

echo "Wrote: $OUT" >&2
echo "Errs:  $ERR" >&2