dehydrate-fs/map.sh

#!/usr/bin/env bash
set -eo pipefail
image="$1"; [ -z "$1" ] && { echo "No image file specified." >&2; exit 1; }

inspect () {
	(	#Jerry-rig error handling onto debugfs
		{ debugfs -R "$*" "$image" 1>&3; } 2>&1 | tail -n +2 |\
		{ if IFS= read -r line; then
			printf '%s\n' "$line"
			cat; exit 1
		  fi; } >&2
	) 3>&1 | cat #Never invoke pager.
}
BLOCKSIZE=`inspect stats | sed -nE 's/^Block size:\s+([0-9]+)$/\1/p'`

#Prefer same filesystem for efficiency on large files
TMPDIR="$(mktemp -d "`pwd`/.pool.XXXXXXXX")" || "$(mktemp -d)" || exit 1
trap 'rm -rf -- "$TMPDIR"' EXIT

size () {
	inspect stat "$1" \
		| head -n3 \
		| sed -nE 's/.*Size: ([0-9]+).*/\1/p'
}

extents () {
	remainder=$((`size $1`%BLOCKSIZE))
	inspect dump_extents "$1" \
		| tail -n +2 \
		| awk '{print $5, $8, $11}' \
		| sed '$s/$/ '$remainder'/'
}

THRESHOLD=${THRESHOLD:-$((1024*1024))}
listall () {
	if [ $# -eq 0 ]; then
		listall "`inspect ls -p /`"
		wait; return
	fi

	perl -p -e 's!/\n$!/\x0!g' <<< "$1" \
	| while IFS=/ read -r -d $'\0' _ inode itype _ _ name isize; do
		case ${itype:0:3} in
		'100') [ $isize -ge $THRESHOLD ] && echo $inode $isize;;
		'040') [[ "${name}" != @(.|..) ]] && {
				echo "Recursing into $name" >&2;
				listall "`inspect ls -p "<$inode>"`" & };;
		esac
	done
}

mkdir -p pool
{
echo -e "#BLOCKSIZE=$BLOCKSIZE\n#INODE SHA256SUM\n#INDEX BLOCK LENGTH [REMAINDER]"
listall \
| while IFS=' ' read -r -d $'\n' inode isize; do
	#TODO: Make asynchronus
	tmp="$(mktemp -p "$TMPDIR")"
	sha=`inspect dump "<$inode>" /dev/stdout \
		| tee "$tmp" \
		| sha256sum \
		| cut -d' ' -f1`
	[ -f pool/$sha ] \
		&& rm -f "$tmp" \
		|| mv -v "$tmp" pool/$sha 1>&2

	echo $inode $sha
	extents "<$inode>" 
done
} > ${2:-/dev/stdout}