截取utf8字符串

string convertString(string str ,int length)

{

vector<std::string > subStrs;

int index=0;

int str_size=str.size();

int bytes=0;

while(index<str_size)

{

bytes=sub_utf_str(str,index,length);

if(bytes==-1)

{

return "wrong";

}

string substr=str.substr(index,bytes);

subStrs.push_back(substr);

index+=bytes;

}

std::string resultStr;

for (unsigned int i = 0;i<subStrs.size();++i)

{

resultStr.append(subStrs[i]).append("\n");

}

return resultStr;

}

//从字符串str的index开始截取length个utf8字符,返回char的个数

int sub_utf_str(string str,int index,int length)

{

int i=0;

int count=0;

int str_size=str.size();

if(index>=str_size)

{

return -1;

}

for(i=0;i<length;i++)

{

char ch=str.at(index);

int bytes=utf8_check_first(ch);

if(bytes==0)

{

return -1;

}

index+=bytes;

count+=bytes;

if(index==str_size)

{

break;

}

if(index>str_size)

{

return -1;

}

}

return count;

}

int utf8_check_first(char byte)

{

unsigned char u = (unsigned char)byte;

if(u&0x80==0)

{

return 1;

}

if(u&0x40==0)

{

return 0;

}

if(u&0x20==0)

{

return 2;

}

if(u&0x10==0)

{

return 3;

}

if(u&0x08==0)

{

return 4;

}

if(u&0x40==0)

{

return 5;

}

if(u&0x20==0)

{

return 6;

}

return 0;

}