Function str_utf8_decode

src/base/str.cpp:952–1017

}

int str_utf8_decode(const char **ptr)
{
	// As per https://encoding.spec.whatwg.org/#utf-8-decoder.
	unsigned char utf8_lower_boundary = 0x80;
	unsigned char utf8_upper_boundary = 0xBF;
	int utf8_code_point = 0;
	int utf8_bytes_seen = 0;
	int utf8_bytes_needed = 0;
	while(true)
	{
		unsigned char byte_value = str_byte_next(ptr);
		if(utf8_bytes_needed == 0)
		{
			if(byte_value <= 0x7F)
			{
				return byte_value;
			}
			else if(0xC2 <= byte_value && byte_value <= 0xDF)
			{
				utf8_bytes_needed = 1;
				utf8_code_point = byte_value - 0xC0;
			}
			else if(0xE0 <= byte_value && byte_value <= 0xEF)
			{
				if(byte_value == 0xE0)
					utf8_lower_boundary = 0xA0;
				if(byte_value == 0xED)
					utf8_upper_boundary = 0x9F;
				utf8_bytes_needed = 2;
				utf8_code_point = byte_value - 0xE0;
			}
			else if(0xF0 <= byte_value && byte_value <= 0xF4)
			{
				if(byte_value == 0xF0)
					utf8_lower_boundary = 0x90;
				if(byte_value == 0xF4)
					utf8_upper_boundary = 0x8F;
				utf8_bytes_needed = 3;
				utf8_code_point = byte_value - 0xF0;
			}
			else
			{
				return -1; // Error.
			}
			utf8_code_point = utf8_code_point << (6 * utf8_bytes_needed);
			continue;
		}
		if(!(utf8_lower_boundary <= byte_value && byte_value <= utf8_upper_boundary))
		{
			// Resetting variables not necessary, will be done when
			// the function is called again.
			str_byte_rewind(ptr);
			return -1;
		}
		utf8_lower_boundary = 0x80;
		utf8_upper_boundary = 0xBF;
		utf8_bytes_seen += 1;
		utf8_code_point = utf8_code_point + ((byte_value - 0x80) << (6 * (utf8_bytes_needed - utf8_bytes_seen)));