#include <stdio.h>
#include <string.h>
#include <stdint.h>

// UTF-8 encoding of 2665
// 11100010 10011001 10100101
        
// 11100010         10011001        10100101
// E   2            9   9           A   5

int utf8_num_bytes(uint8_t first_byte);

int main(void){ 
    // \u only works for code points with up to 4 hex digits
    printf("\u2665\n");
    
    printf("\xE2\x99\xA5\n");
    char *s = "\u2665 beats";
    printf("%s\n",s);
    printf("♥ beats\n"); //ctrl-shift-u 2665
    
    
    //string functions DO NOT WORK WITH UNICODE
    printf("%lu\n",strlen("\u2665")); //3 not 1
    printf("%lu\n",strlen(s));        //9 not 7
  
    //using our own function to work out the number
    //of bytes in unicode character with a given first byte
    uint8_t first_byte = 0xE2;
    printf("Expected nbytes for \\x%hhX is %d\n",first_byte, utf8_num_bytes(first_byte));
    
    return 0;
}


int utf8_num_bytes(uint8_t first_byte){
    if ((first_byte & 0x80) == 0) {
        return 1;
    } else if ((first_byte & 0xe0) == 0xc0) {
        return 2;
    } else if ((first_byte & 0xf0) == 0xe0) {
        return 3;
    } else if ((first_byte & 0xf8) == 0xf0) {
        return 4;
    } else {
        fprintf(stderr,"Not a valid first byte\n");
        return 0;
    }
}