Function Utf8ValidateImpl

internal/utf8.cc:277–314

template <typename BufferedByteReader>
std::pair<size_t, bool> Utf8ValidateImpl(BufferedByteReader* reader) {
  size_t count = 0;
  while (reader->HasRemaining()) {
    const auto b = static_cast<uint8_t>(reader->Read());
    if (b < kUtf8RuneSelf) {
      count++;
      continue;
    }
    const auto leading = kLeading[b];
    if (leading == kXX) {
      return {count, false};
    }
    const auto size = static_cast<size_t>(leading & 7) - 1;
    if (size > reader->Remaining()) {
      return {count, false};
    }
    const absl::string_view segment = reader->Peek(size);
    const auto& accept = kAccept[leading >> 4];
    if (static_cast<uint8_t>(segment[0]) < accept.first ||
        static_cast<uint8_t>(segment[0]) > accept.second) {
      return {count, false};
    } else if (size == 1) {
      count++;
    } else if (static_cast<uint8_t>(segment[1]) < kLow ||
               static_cast<uint8_t>(segment[1]) > kHigh) {
      return {count, false};
    } else if (size == 2) {
      count++;
    } else if (static_cast<uint8_t>(segment[2]) < kLow ||
               static_cast<uint8_t>(segment[2]) > kHigh) {
      return {count, false};
    } else {
      count++;
    }
    reader->Advance(size);
  }
  return {count, true};
}

}  // namespace