Interesting approach. It seems to be more efficient comparing to keep-history way. Especially for huge sets. It consumes no memory and complexity is almost linear.
Thanks for the inspiration.

Nice one. Thanks.
Would be better suited in a class, I guess.
And automatic move could be next kata/method :)

The same case. It seems to be a typo in the test definition.

• As ptya writes encoded text can't be lenger than input
• Encoded text, marked as failed result, can be reversibly decoded

To encode: どもあみがとございまぬ
Encoded: どしららがろござやまぉ
ERR:Expected: どしららがろござや���ぉ
To decode: どしららがろござやまぉ
Decoded: どもあみがとございまぬ

Nice kata, thanks :)