Converting bytes16 to int128

Question

I am working on a little decentralized game with a world made of plots. It has a map which links a location to a Plot struct. A location is saved as a bytes16 but represents two int64 (coordinates can be negative and the center is in 0, 0). How can I convert them to int64 (it seems I can only convert them to uint64)?

Here is what I tried:

    function convertToCoordinates(bytes32 input)
        public
        pure
        returns (uint128 x, uint128 y)
    {
        bytes16 xbytes;
        bytes16 ybytes;
        (xbytes, ybytes) = split(input);

        x = int128(int128(uint128(xbytes)) - 2**127 - 1);
        y = int128(int128(uint128(ybytes)) - 2**127 - 1);
    }

Answer 1

Here is a solution:

    function split(bytes32 source) public pure returns (bytes16 x, bytes16 y) {
        bytes16[2] memory output = [bytes16(0), 0];
        assembly {
            mstore(output, source)
            mstore(add(output, 16), source)
        }
        return (output[0], output[1]);
    }

    function convertToCoordinates(bytes32 input)
        public
        pure
        returns (int128 x, int128 y)
    {
        bytes16 xbytes;
        bytes16 ybytes;
        (xbytes, ybytes) = split(input);

        uint128 ux = uint128(xbytes);
        uint128 uy = uint128(ybytes);

        x = int128(ux);
        y = int128(uy);
    }

And since one is not very useful without the other, here is a way to do the opposite:

    function merge(bytes16 sourceA, bytes16 sourceB)
        public
        pure
        returns (bytes32)
    {
        return bytes32((uint256(uint128(sourceA)) << 128) | uint128(sourceB));
    }

    function convertFromCoordinates(int128 x, int128 y)
        public
        pure
        returns (bytes32 output)
    {
        return merge(bytes16(uint128(x)), bytes16(uint128(y)));
    }