Function GetNextCodePointAndAdvance

lib/yamlcpp/src/emitterutils.cpp:90–132

bool IsTrailingByte(char ch) { return (ch & 0xC0) == 0x80; }

bool GetNextCodePointAndAdvance(int& codePoint,
                                std::string::const_iterator& first,
                                std::string::const_iterator last) {
  if (first == last)
    return false;

  int nBytes = Utf8BytesIndicated(*first);
  if (nBytes < 1) {
    // Bad lead byte
    ++first;
    codePoint = REPLACEMENT_CHARACTER;
    return true;
  }

  if (nBytes == 1) {
    codePoint = *first++;
    return true;
  }

  // Gather bits from trailing bytes
  codePoint = static_cast<unsigned char>(*first) & ~(0xFF << (7 - nBytes));
  ++first;
  --nBytes;
  for (; nBytes > 0; ++first, --nBytes) {
    if ((first == last) || !IsTrailingByte(*first)) {
      codePoint = REPLACEMENT_CHARACTER;
      break;
    }
    codePoint <<= 6;
    codePoint |= *first & 0x3F;
  }

  // Check for illegal code points
  if (codePoint > 0x10FFFF)
    codePoint = REPLACEMENT_CHARACTER;
  else if (codePoint >= 0xD800 && codePoint <= 0xDFFF)
    codePoint = REPLACEMENT_CHARACTER;
  else if ((codePoint & 0xFFFE) == 0xFFFE)
    codePoint = REPLACEMENT_CHARACTER;
  else if (codePoint >= 0xFDD0 && codePoint <= 0xFDEF)
    codePoint = REPLACEMENT_CHARACTER;
  return true;
}

void WriteCodePoint(ostream_wrapper& out, int codePoint) {
  if (codePoint < 0 || codePoint > 0x10FFFF) {