最近项目中用到需要给出每一个字在string中的索引,但是又因为中文字符跟英文字符长度不一样,得分开处理,

在这里记录一下。

想要达到的效果如下:

将 “测试3.1415engEng”分割开

c++读取utf-8格式中英文混合string-LMLPHP

代码:

std::vector <std::string> splitEachChar(const string chars)
{
std::vector<std::string> words;
std::string input(chars);
int len = input.length();
int i = ; while (i < len) {
assert ((input[i] & 0xF8) <= 0xF0);
int next = ;
if ((input[i] & 0x80) == 0x00) {
std::cout << "one character: " << input[i] << std::endl;
} else if ((input[i] & 0xE0) == 0xC0) {
next = ;
std::cout << "two character: " << input.substr(i, next) << std::endl;
} else if ((input[i] & 0xF0) == 0xE0) {
next = ;
std::cout << "three character: " << input.substr(i, next) << std::endl;
} else if ((input[i] & 0xF8) == 0xF0) {
next = ;
std::cout << "four character: " << input.substr(i, next) << std::endl;
}
words.push_back(input.substr(i, next));
i += next;
}
return words;
}
void testtemp()
{
string input;
while ()
{
getline(cin,input);
if(input == "exit") break;
cout<<"--------------------------------"<<endl;
vector <std::string> ret = splitEachChar(input); cout<<input<<endl;
for(auto it : ret)cout<<it<<endl;
cout<<"--------------------------------"<<endl;
}
}
int main()
{
testtemp();
return ;
}

参考:

https://blog.csdn.net/cy_tec/article/details/87884177

05-26 16:49
查看更多