std::mblen

#include <clocale>
#include <string>
#include <iostream>
#include <cstdlib>
#include <stdexcept>
 
// the number of characters in a multibyte string is the sum of mblen()'s
// note: the simpler approach is std::mbstowcs(NULL, s.c_str(), s.size())
std::size_t strlen_mb(const std::string& s)
{
    std::size_t result = 0;
    const char* ptr = &s[0];
    const char* end = ptr + s.size();
    std::mblen(NULL, 0); // reset the conversion state
    while (ptr < end) {
        int next = std::mblen(ptr, end-ptr);
        if (next == -1) {
            throw std::runtime_error("strlen_mb(): conversion error");
        }
        ptr += next;
        ++result;
    }
    return result;
}
 
int main()
{
    // allow mblen() to work with UTF-8 multibyte encoding
    std::setlocale(LC_ALL, "en_US.utf8");
    // UTF-8 narrow multibyte encoding
    std::string str = u8"z\u00df\u6c34\U0001d10b"; // or u8"zß水𝄋"
                      // or "\x7a\xc3\x9f\xe6\xb0\xb4\xf0\x9d\x84\x8b";
    std::cout << str << " is " << str.size() << " bytes, but only "
              << strlen_mb(str) << " characters\n";
}