UTF_stuff.d

BCBuilitin_NextUTF8CharT nextChar = (const Heap* heapPtr, uint* argPtr) pure {
    uint stringPos = *argPtr;
    uint stringPosOverFour = stringPos / 4;
    uint bytePos = stringPos % 4;
    uint loWord = heapPtr._heap[stringPosOverFour];

    uint c = intrin_Byte3(loWord, bytePos);
    if (!((cast(ubyte)c) & 0x80)) {
        (*argPtr) = stringPos + 1;
        return BCValue(Imm32(c));
    } else {
        //TODO check for invalid sequences!
        if ((c & 0b0010_0000) == 0) { // we have a 2byte sequence
            if (bytePos++ == 3) { // probability for this is 1/4;
                c |= (intrin_Byte3(heapPtr._heap[++stringPosOverFour], 0) & 0x80) >> 5;
            } else {
                c |= (intrin_Byte3(loWord, bytePos) & 0x80) >> 5; 
            }
            (*argPtr) = stringPos + 2;
        } else if ((c & 0b0001_0000) == 0) { // we have a 3 byte sequence
            if (bytePos++ == 3) { // probability for this is 1/4;
                uint hiWord = heapPtr._heap[++stringPosOverFour];
                c |= (intrin_Byte3(hiWord, 0) & 0x80) >> 4;
                c |= (intrin_Byte3(hiWord, 1) & 0x80) >> (4+6);
            } else {
                c |= (intrin_Byte3(loWord, bytePos++) & 0x7f) >> 4;
                if (bytePos++ == 3) { // the probabilty for this is 2/4
                    c |= (intrin_Byte3(heapPtr._heap[++stringPosOverFour], 0) & 0x7f) >> (4+6);
                }
            }
            (*argPtr) = stringPos + 3;
        } else {
            //TODO when we support more the 3 chars cache the hi-word;
            assert(0, "For now we don't support UTF8 Chars bigger then 3");
        }

        return(BCValue(Imm32(c)));
    }
};