utf8文字数カウント

utf8count.c

#include <stdio.h>

int byte_size(const unsigned char c) {
  if(c <= 0x7f) {
    return 1;
  } else if(0xc2 <= c && c <= 0xdf) {
    return 2;
  } else if(0xe0 <= c && c <= 0xef) {
    return 3;
  } else if(0xf0 <= c && c <= 0xff) {
    return 4;
  } else {
    return -1;
  }
}

int length(const unsigned char s[], const int size) {
  int count = 0;
  int i = 0;

  for(i = 0; i < size;) {
    const int bytes = byte_size(s[i]);
    if(bytes < 0) {
      break;
    } else {
      count++;
      i+= bytes;
    }
  }
  return count -1;
}

int main() {
  unsigned char s[] = "あいうえお鬼の葛餅 🔣✔";
  int l = length(s, sizeof(s));
  printf("length(%s) = %d\n", s, l);
  return 0;
}