dflatten

#!/bin/bash

function containsElement () {
  local e
  for e in "${@:2}"; do [[ "$e" == "$1" ]] && return 0; done
  return 1
}

function findImages () {
  jq $1 /var/lib/docker/repositories-aufs | sed "s/\"//g"
}

function readLayers () {
  local arr
  readarray -t arr < <(cat /var/lib/docker/aufs/layers/$1)

  local farr=("$1")
  farr+=("${arr[@]}")
  echo ${farr[@]}
}

function mountLayers () {
  #declare -a layers=("${!1}")
  for i in "${!1}"
  do
    mount -o remount,append:"/var/lib/docker/aufs/diff/$i/"=ro+wh $2
  done	
}

while [[ $# > 1 ]]
do
	key="$1"
	shift

	case $key in
	    -o|--original)
        ORIGINAL_ARG="$1"
	    shift
	    ;;
	    -t|--target)
	    TARGET_ARG="$1"
	    shift
	    ;;
	    *)
           # unknown option
	    ;;
	esac
done

if [ -z "$ORIGINAL_ARG" ]; then
	echo "$(basename "$0") -o original_image_name -t target_image_name "
	exit 1
fi

arrIN=(${ORIGINAL_ARG//:/ })
ORIGINAL_NAME=${arrIN[0]}
ORIGINAL_TAG=${arrIN[1]:=latest}

arrIN=(${TARGET_ARG//:/ })
TARGET_NAME=${arrIN[0]:=${ORIGINAL_NAME}_flatten}
TARGET_TAG=${arrIN[1]:=$ORIGINAL_TAG}

TEMP_DIR=`mktemp -d`

### Find tagged images
readarray -t TAGGED_IMAGES < <(findImages '.["Repositories"]|.[]|.[]')

### Find original image ID
ORIGINAL_ID=`findImages '.["Repositories"]|.["'$ORIGINAL_NAME'"]|.["'$ORIGINAL_TAG'"]'`

if [ "null" = "$ORIGINAL_ID" ]; then
	echo "Can't find image $ORIGINAL_NAME:$ORIGINAL_TAG"
	exit 1
fi

### Target ID is reverse of original ID, so its stable
TARGET_ID=`echo $ORIGINAL_ID | rev`

### Do we have image in cache?
if [ -f "/var/lib/docker/aufs/layers/"$TARGET_ID ]; then
   echo Using cache
   echo $TARGET_ID
   docker tag $TARGET_ID $TARGET_NAME:$TARGET_TAG
   exit 0
else
   echo Flatten image to $TARGET_ID
fi

### Find all layers of original image
ORIGINAL_LAYERS=( $(readLayers $ORIGINAL_ID) )
ORIGINAL_PARENT_ID=${ORIGINAL_LAYERS[1]}

### Find parent ID
for i in "${ORIGINAL_LAYERS[@]:1}"
do
	containsElement $i "${TAGGED_IMAGES[@]}"
	if [ "$?" == "0" ]
	then
		PARENT_ID=$i
		break
	fi
done

### Find parent images
PARENT_LAYERS=( $(readLayers $PARENT_ID) )

### Make temp folders
ORIGINAL_DIR=$TEMP_DIR/source
FLATTEN_DIR=$TEMP_DIR/flatten

mkdir $FLATTEN_DIR
mkdir $ORIGINAL_DIR
mkdir $TEMP_DIR/empty
mkdir $TEMP_DIR/diff

### Recovery original image
mount -t aufs -o ro,br:$TEMP_DIR/empty none $ORIGINAL_DIR
mountLayers ORIGINAL_LAYERS[@] $ORIGINAL_DIR

### Build diff
mount -t aufs -o rw,br:$TEMP_DIR/diff/=rw none $FLATTEN_DIR
mountLayers PARENT_LAYERS[@] $FLATTEN_DIR

rsync -aAXvq --exclude={"/dev/*","/proc/*","/sys/*","/tmp/*","/run/*","/mnt/*","/media/*","/lost+found"} --delete "$ORIGINAL_DIR/" $FLATTEN_DIR

umount $FLATTEN_DIR
umount $ORIGINAL_DIR

### Build tar for load
LL_DIR=$TEMP_DIR/result/$TARGET_ID

mkdir -p $LL_DIR
tar -cpf $LL_DIR/layer.tar --numeric-owner -C $TEMP_DIR/diff .
echo 1.0 > $LL_DIR/VERSION
cat /var/lib/docker/graph/$ORIGINAL_ID/json | sed s/$ORIGINAL_ID/$TARGET_ID/g | sed s/$ORIGINAL_PARENT_ID/$PARENT_ID/g > $LL_DIR/json
echo '{"'$TARGET_NAME'":{"'$TARGET_TAG'":"'$TARGET_ID'"}}' > $TEMP_DIR/result/repositories

tar -cpf $TEMP_DIR/tmp.tar --numeric-owner -C $TEMP_DIR/result .

docker load < $TEMP_DIR/tmp.tar

rm -r $TEMP_DIR