utf16.c

#include "utf16.h"
#include <errno.h>

uint8_t
isutf16(const unsigned char* string)
{
  uint16_t W1, W2;
  unsigned char *ptr = (unsigned char*) string;
  while((W1 = *((uint16_t*) ptr)) != 0x0L) {
    if(W1 < 0xd800 || W1 > 0xdfff) {
      ptr += 2;
      continue;
    }

    if(W1 >= 0xd800 && W1 <= 0xdbff) {
      W2 = *((uint16_t*) (ptr + 2));
      if(W2 >= 0xdc00 && W2 <= 0xdfff) {
        ptr += 4;
        continue;
      }
    }

    errno = EINVAL;
    return 0;
  }

  return 1;
}

int32_t
utf16len(const unsigned char* string)
{
  uint16_t W1, W2;
  int len = 0;
  unsigned char *ptr = (unsigned char*) string;
  while((W1 = *((uint16_t*) ptr)) != 0x0L) {
    if(W1 < 0xd800 || W1 > 0xdfff) {
      len++; ptr += 2;
      continue;
    }

    if(W1 >= 0xd800 && W1 <= 0xdbff) {
      W2 = *((uint16_t*) (ptr + 2));
      if(W2 >= 0xdc00 && W2 <= 0xdfff) {
        len++; ptr += 4;
        continue;
      }
    }

    errno = EINVAL;
    return -1;
  }

  return len;
}

utf16.h

#ifndef _UTF_16_H
#define _UTF_16_H

#include <stdint.h>

/**
 *
 */
uint8_t isutf16(const unsigned char*);

/**
 *
 */
int32_t utf16len(const unsigned char*);

#endif