Function char2Byte

src/python/handler/py_main.cpp:19–54

#endif

void char2Byte(const char* strBegin, const char* strEnd, vector<uint32_t>& startPos, vector<uint16_t>& length)
{
	if (strBegin == strEnd) return;
	vector<size_t> charPos;
	auto it = strBegin;
	for (; it != strEnd; )
	{
		charPos.emplace_back(it - strBegin);
		uint8_t c = *it;
		if ((c & 0xF8) == 0xF0)
		{
			it += 4;
		}
		else if ((c & 0xF0) == 0xE0)
		{
			it += 3;
		}
		else if ((c & 0xE0) == 0xC0)
		{
			it += 2;
		}
		else if ((c & 0x80))
		{
			throw std::runtime_error{ "utf-8 decoding error" };
		}
		else it += 1;
	}
	charPos.emplace_back(strEnd - strBegin);

	for (size_t i = 0; i < startPos.size(); ++i)
	{
		size_t s = startPos[i], e = (size_t)startPos[i] + length[i];
		startPos[i] = charPos[s];
		length[i] = charPos[e] - charPos[s];
	}
}

void char2Byte(const string& str, vector<uint32_t>& startPos, vector<uint16_t>& length)
{