按一定长度截取utf-8编码的字符串

最新推荐文章于 2022-05-19 10:18:29 发布

hddygs001

最新推荐文章于 2022-05-19 10:18:29 发布

阅读量2k

点赞数 1

本文链接：https://blog.csdn.net/hddygs001/article/details/20133199

版权

有关学习专栏收录该内容

79 篇文章 0 订阅

订阅专栏

int getChLength(unsigned char c)
{

	if (c < 0x7F) 
	{
		return 1;
	}else if ((c & 0xE0) == 0xC0)
	{
		return 2;
	}else if ((c & 0xF0) == 0xE0)
	{
		return 3;
	}else if ((c & 0xF8) == 0xF0)
	{
		return 4;
	}else if ((c & 0xFC) == 0xF8)
	{
		return 5;
	}else if ((c & 0xFE) == 0xFC)
	{
		return 6;
	}else
		return -1;
}

std::string sUtf8Substr(std::string sFormerStr, int iRsltLen)
{
	if (sFormerStr.size() <= iRsltLen)
	{
		return sFormerStr;
	}

	char* cbuf  = new char[iRsltLen];
	memset(cbuf, 0, iRsltLen);
	int i = 0;
	int j = 0;
	while (i < iRsltLen)
	{
		j = getChLength(sFormerStr[i]);

		if (j == -1)
			break;

		strcat(cbuf, sFormerStr.substr(i, j).c_str());
		i += j;
	}

	std::string sTmp(cbuf);
	delete [] cbuf;

	if (sTmp.size() > iRsltLen)
	{
		sTmp.erase(sTmp.size() - 1 - j, j);
	}

	return sTmp;
}