thx man for your reply, I was stupid too with this 26 symbols

im just simply adding the index of PW str am I doing it wrong ?

I am not sure if I understand you, but I believe this is your problem.
The katakana character set used by this kata is not a consistent sequence of codepoints, and you need to use actual definition of the alphabet to check what a character shifted for x positions is. You cannot just convert a charater to its code point, add x to it, and convert it back to a character - this wont work. For the Latin alfabet, it's true that 'c' + 3 == 'f', but it's a coincidence which does not hapen for the katakana alphabet in this kata.

is that means PW is カタカナ and is should decode or encode カタカナ to タモタワ?

Yes, you can log stuff to console to see this better.

coz theres no way Japanese used the abc just like English right?

It should be handled the same way, I think.