2.3 多国语言编码范围

/*
  功能：  UTF-8编码转UTF-16编码
  pUTF8： UTF-8编码数据
  pUTF16：UTF-16编码数据
  return：UTF-16字符的字节长度
*/
unsigned char UTF8toUTF16(unsigned char *pUTF8, unsigned short *pUTF16)
{
  unsigned char bytes[3];
  unsigned short unicode16;

  bytes[0] = *pUTF8++;
  if (bytes[0] < 0x80) {
    *pUTF16 = bytes[0];
    return 1;
  }

  bytes[1] = *pUTF8++;
  if (bytes[0] >= 0xC0 && bytes[0] < 0xE0) {
    // 2字节 UTF-8
    unicode16  = (bytes[0] & 0x1F) << 6;
    unicode16 |= (bytes[1] & 0x3F);
    *pUTF16 = unicode16;
    return 2;
  }

  bytes[2] = *pUTF8++;
  if (bytes[0] >= 0xE0 && bytes[0] < 0xF0) {
    // 3字节 UTF-8
    unicode16  = (bytes[0] & 0x0F) << 12;
    unicode16 |= (bytes[1] & 0x3F) << 6;
    unicode16 |= (bytes[2] & 0x3F);
    *pUTF16 = unicode16;
    return 3;
  }

  return 0;
}