Fix regression in GetVarStr introduced in commit '8a5ada2d7'

[peolsson]

When calling GetVarStr with a NULL pointer, it was originally expected that the function returned the length of the string, even though the string itself could not be stored. This is also what is expected in N2kDeviceList.cpp, which makes the change in '8a5ada2d7' to break the expected behaviour.

This change restores the original behaviour.

[ttlappalainen]

You need to check length for possible UTF8:

...
//*****************************************************************************
int N2kUCS2ToUTF8Len(const unsigned char *str, int strLen, unsigned char nulChar=0xff) {
  int UTF8Len=0;

  for (int i=0; i+1<strLen; i+=2) {
    uint16_t UCS2Char = str[i] + (str[i+1] << 8);

    // Decode
    if ( 0x80 > UCS2Char ) {
      //Tokensize: 1 byte
      if ( UCS2Char!=nulChar ) UTF8Len++;
    } else if ( 0x800 > UCS2Char ) {
      //Tokensize: 2 bytes
      UTF8Len+=2;
    } else {
      //Tokensize: 3 bytes
      UTF8Len+=3;
    }
  }

  return UTF8Len;
}

//*****************************************************************************
int N2kUCS2ToUTF8(const unsigned char *str, int strLen, char* buf, int bufLen, unsigned char nulChar=0xff) {
...
    if ( Type==0x01 ) {
      StrBufSize=Len;
    } else {
      const unsigned char *UnicodeStr=&Data[Index];
      StrBufSize=N2kUCS2ToUTF8Len(UnicodeStr,Len,nulChar);
    }
...

[peolsson]

Thanks for the update. I added handling for unicode strings now as well.

[ttlappalainen]

Move N2kUCS2ToUTF8Len above N2kUCS2ToUTF8 as I pointed. They have nearley same functionality so better to keep them near each other.