UTF8 encode

utf8.cpp

#include <iostream>

using namespace std;
/*
0xxxxxxx A single-byte US-ASCII code (from the first 127 characters)
110xxxxx One more byte follows
1110xxxx Two more bytes follow
11110xxx Three more bytes follow

10xxxxxx A continuation of one of the multi-byte characters
*/
bool validUTF8(const char *s) {
    int m, r, i = 0;
    while (*s != '\0') {
        m = 7;
        while (m > 0 && (*s & (1 << m)) > 0) --m;
        ++s;
        r = 6-m;
        if (r == -1) // single byte
            continue;
        else if (r == 0 || r > 3) // continuation or bytes more than 3
            return false;
        else {
            while (r > 0 && *s != '\0') {
                if ((*s & (1<<7)) && (*s & (1<<6)) == 0) {
                    ++s;
                    --r;
                } else {
                    break;
                }
            }
            if (r > 0) return false;
        }
    }
    return true;
}

// 11010000
// 10011000
int main() {
    const char *s= "Hello, 李者"; 
    cout<<validUTF8(s)<<endl;
}