bool valid_utf8()

in crawl/filter_utf8.cc [62:94]

• 29 lines of code
• 17 McCabe index (conditional complexity)

bool valid_utf8(uint8_t* str, size_t length)
{
  uint8_t* end = str + length;
  while (str < end) {
    if (str[0] < 0x80) {
      // 0.xxxxxxx
      str += 1;
    } else if ((str[0] & 0xe0) == 0xc0) {
      // 110.xxxxx 10.xxxxxx
      if (str + 1 >= end) return false;
      if (!continuation(str + 1, 1)) return false;
      if (overlong_2(str)) return false;
      str += 2;
    } else if ((str[0] & 0xf0) == 0xe0) {
      // 1110.xxxx 10.xxxxxx 10.xxxxxx
      if (str + 2 >= end) return false;
      if (!continuation(str + 1, 2)) return false;
      if (overlong_3(str)) return false;
      if (surrogate(str)) return false;
      str += 3;
    } else if ((str[0] & 0xf8) == 0xf0) {
      // 11110.xxx 10.xxxxxx 10.xxxxxx 10.xxxxxx
      if (str + 3 >= end) return false;
      if (!continuation(str + 1, 3)) return false;
      if (overlong_4(str)) return false;
      if (invalid(str)) return false;
      str += 4;
    } else {
      return false;
    }
  }
  return true;
}