utf8.c

static inline const uint8_t *DecodeOneUTF8(const uint8_t *utf8, uint32_t *outCodepoint) {
    uint8_t byteOne = utf8[0];
    uint8_t byteTwo = utf8[1];
    uint8_t byteThree = utf8[2];
    uint8_t byteFour = utf8[3];
    
    uint8_t bit1 = byteOne >> 7;
//    uint8_t bit2 = (byteOne >> 6) & 1;
    uint8_t bit3 = (byteOne >> 5) & 1;
    uint8_t bit4 = (byteOne >> 4) & 1;
//    uint8_t bit5 = (byteOne >> 3) & 1;
    
    uint8_t hasTwo = bit1;
    uint8_t hasThree = hasTwo & bit3;
    uint8_t hasFour = hasThree & bit4;
    
    uint8_t byteOneBits = 7 - hasTwo - hasTwo - hasThree - hasFour;
    uint8_t byteOneMask = 0xff >> (8 - byteOneBits);
    
    uint32_t codepoint = byteOne & byteOneMask;
    
    uint8_t trailingMask = 0x3f;
    
    codepoint <<= hasTwo * 6;
    codepoint |= hasTwo * (byteTwo & trailingMask);
    
    codepoint <<= hasThree * 6;
    codepoint |= hasThree * (byteThree & trailingMask);
    
    codepoint <<= hasFour * 6;
    codepoint |= hasFour * (byteFour & trailingMask);
    
    *outCodepoint = codepoint;
    
    return utf8 + 1 + hasTwo + hasThree + hasFour;
}

size_t DecodeUTF8(const uint8_t *utf8, uint32_t *destination) {
    size_t length = 0;
    while(*utf8 != '\0') {
        utf8 = DecodeOneUTF8(utf8, destination);
        destination++;
        length++;
    }
    return length;
}