dedup.sh

#!/bin/sh

set -Cefu -o pipefail

cleanup() {
	if [ -d "${TMP_DIR:?}" ]; then
		echo rm -rf -- "${TMP_DIR:?}"
	fi
	trap - EXIT
}
: "${BLOB_DIR:=/var/db/blobs}"
[ -d "${BLOB_DIR:?}" ] || mkdir -p "${BLOB_DIR:?}"
readonly BLOB_DIR TMP_DIR="${BLOB_DIR:?}/.tmp.$$"
trap cleanup EXIT INT PIPE ALRM TERM XCPU XFSZ VTALRM PROF USR1 USR2

TAG='' SRC=''
while getopts 't:s:' ARG; do
	case "$ARG" in
		t) TAG="$OPTARG" ;;
		s) SRC="$OPTARG" ;;
		*) exit 64;;
	esac
done
readonly TAG SRC TAG_DIR="${BLOB_DIR:?}/$TAG"

case "$TAG" in
	'')            echo "Tag (-t <tag>) is missing (or empty)."                       >&2; exit 64;;
	.*)            echo "Tags must not start with a dot."                             >&2; exit 64;;
	*[$'/ \t\n']*) echo "Tags must not contain a slashes, spaces, tabs or newlines."  >&2; exit 64;;
esac

if [ -z "$SRC" ]; then
	echo "Source (-s <source>) is missing." >&2; exit 64
elif [ ! -d "$SRC" ]; then
	echo "The source must be an existing directory." >&2; exit 64
fi

if [ -d "${TAG_DIR:?}" ]; then
	echo "The tag already exists."
	exit 0
fi

rm -rf -- "${TMP_DIR:?}/.tmp.$$"
mkdir -p "${TMP_DIR:?}"
cp -an -- "${SRC:?}/" "${TMP_DIR:?}/content"
cd "${TMP_DIR:?}/content"
umask 377
mtree -c -n -K flags,gid,mode,sha512,size,time,type,uid >../mtree
umask 022
IFS=' '
<../mtree mtree -D -R type,uid,gid,mode,nlink,time,size -K sha512,flags | while read -r HASH FLAGS FILE; do
	if [ -z "$FILE" ] || [ -L "$FILE" ] || [ ! -f "$FILE" ] || [ ! -s "$FILE" ]; then
		continue
	fi
	HASH="${HASH#sha512=}"
	FLAGS="${FLAGS#flags=}"
	DIR_XX="${HASH%??????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????}"
	DIR_XXXX="${HASH%????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????}"
	INFIX=".blobs/${DIR_XX:?}/${DIR_XXXX:?}"
	[ -d "../../${INFIX:?}" ] || mkdir -p "../../${INFIX:?}"
	[ -s "../../${INFIX:?}/${HASH:?}" ] || install -S -C -m 444 -f uarch -- "$FILE" "../../${INFIX:?}/${HASH:?}"
	[ -d "../${INFIX:?}" ] || mkdir -p "../${INFIX:?}"
	[ -s "../${INFIX:?}/${HASH:?}" ] || ln -hn -- "../../${INFIX:?}/${HASH:?}" "../${INFIX:?}/${HASH:?}"
	case ",$FLAGS," in
		*,schg,*)
			chflags -- noschg "$FILE"
			chmod -- u+w "$FILE"
			cat "../${INFIX:?}/${HASH:?}" >| "$FILE"
			chmod -- u-w "$FILE"
			chflags -- schg "$FILE"
			;;
		*)
			cat "../${INFIX:?}/${HASH:?}" >| "$FILE"
			;;
	esac
done
cd -
mv -nh -- "${TMP_DIR:?}" "${TAG_DIR:?}"