strnlen for UTF-8. Considers the character boundary and an invalid sequence.

strnlen_utf8.c

/**
 * @file
 * strnlen for UTF-8. Considers the character boundary and an invalid sequence.
 */

#include <stdlib.h>

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

/**
 * Returns the length of the given null-terminated byte string.
 * @param str pointer to the null-terminated byte string to be examined.
 * @param maxlen maximum number of characters to examine.
 * @return The length of the null-terminated byte string str on success, zero if str is an invalid utf8 string or NULL, maxlen if the null character was not found.
 */
size_t strnlen_utf8(const char *str, size_t maxlen) {
  size_t len = 0;

  if (str == NULL) {
    return 0;
  }

  /* Calculate the length of the string. */
  while (len < maxlen && str[len] != '\0') {
    unsigned char tempchar = (unsigned char)str[len];
    size_t charsize;
    size_t off;

    if (tempchar <= 0x7F) {
      /* 1 byte character. */
      charsize = 1;
    }
    else if ((tempchar & 0xE0) == 0xC0) {
      /* 2 bytes character. */
      charsize = 2;
    }
    else if ((tempchar & 0xF0) == 0xE0) {
      /* 3 bytes character. */
      charsize = 3;
    }
    else if ((tempchar & 0xF8) == 0xF0) {
      /* 4 bytes character. */
      charsize = 4;
    }
    else if ((tempchar & 0xFC) == 0xF8) {
      /* 5 bytes character. */
      charsize = 5;
    }
    else if ((tempchar & 0xFE) == 0xFC) {
      /* 6 bytes character. */
      charsize = 6;
    }
    else {
      /* Character out of range. */
      return 0;
    }

    /* Check trailer bytes. */
    for (off = 1; off < charsize; off++) {
      if (((unsigned char)str[len + off] & 0xC0) != 0x80) {
        /* Invalid bit pattern. */
        return 0;
      }
    }

    /* Last character? */
    if (len + charsize > maxlen) {
      return len;
    }

    /* Otherwise, go next. */
    len += charsize;
  }

  /* Return the length. */
  return (len <= maxlen) ? len : maxlen;
}

int main(void) {
  unsigned char s[] = {
    0xE3, 0x81, 0x93, 0xE3, 0x82, 0x93, 0xE3, 0x81, 0xAB, 0xE3, 0x81, 0xA1, 0xE3, 0x81, 0xAF, 0
  };

  printf("%u\n", strnlen_utf8((char *)s, 14));
  return 0;
}