UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_iterate()

in subversion/libsvn_subr/utf8proc/utf8proc.c [113:159]

• 40 lines of code
• 23 McCabe index (conditional complexity)

UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_iterate(
  const utf8proc_uint8_t *str, utf8proc_ssize_t strlen, utf8proc_int32_t *dst
) {
  utf8proc_uint32_t uc;
  const utf8proc_uint8_t *end;

  *dst = -1;
  if (!strlen) return 0;
  end = str + ((strlen < 0) ? 4 : strlen);
  uc = *str++;
  if (uc < 0x80) {
    *dst = uc;
    return 1;
  }
  /* Must be between 0xc2 and 0xf4 inclusive to be valid */
  if ((uc - 0xc2) > (0xf4-0xc2)) return UTF8PROC_ERROR_INVALIDUTF8;
  if (uc < 0xe0) {         /* 2-byte sequence */
     /* Must have valid continuation character */
     if (str >= end || !utf_cont(*str)) return UTF8PROC_ERROR_INVALIDUTF8;
     *dst = ((uc & 0x1f)<<6) | (*str & 0x3f);
     return 2;
  }
  if (uc < 0xf0) {        /* 3-byte sequence */
     if ((str + 1 >= end) || !utf_cont(*str) || !utf_cont(str[1]))
        return UTF8PROC_ERROR_INVALIDUTF8;
     /* Check for surrogate chars */
     if (uc == 0xed && *str > 0x9f)
         return UTF8PROC_ERROR_INVALIDUTF8;
     uc = ((uc & 0xf)<<12) | ((*str & 0x3f)<<6) | (str[1] & 0x3f);
     if (uc < 0x800)
         return UTF8PROC_ERROR_INVALIDUTF8;
     *dst = uc;
     return 3;
  }
  /* 4-byte sequence
     Must have 3 valid continuation characters */
  if ((str + 2 >= end) || !utf_cont(*str) || !utf_cont(str[1]) || !utf_cont(str[2]))
     return UTF8PROC_ERROR_INVALIDUTF8;
  /* Make sure in correct range (0x10000 - 0x10ffff) */
  if (uc == 0xf0) {
    if (*str < 0x90) return UTF8PROC_ERROR_INVALIDUTF8;
  } else if (uc == 0xf4) {
    if (*str > 0x8f) return UTF8PROC_ERROR_INVALIDUTF8;
  }
  *dst = ((uc & 7)<<18) | ((*str & 0x3f)<<12) | ((str[1] & 0x3f)<<6) | (str[2] & 0x3f);
  return 4;
}