@wech71 thanks 👍
I move method into other class to simplify test and fixed code style with autoformatter.
Although, I was not able to trigger branch which decodes 6-byte char, looks like my sample chars was splitted into two 3-byte chars. It is not urgent, so we may add a test for that case later.

I apologize if this is incorrect now as you said, I should not merge to the master branch

You do everything right. I was trying to say, that branch in your fork shouldn't be a master. Because on Github master branch is write protected, and I will not be able to add fixes commit, like I do.

thanks for accepting my changes :-)

looks like my sample chars was splitted into two 3-byte chars.

hm, I just had another look at the spec and it seems I made a small mistake. The check for 6-byte encoding according 4.12. must be done before 3-byte encoding 4.8, because both use the same upper nibble of first byte and the same two bits for byte 2 and 3.

This way the code currently always incorrectly enters 3-byte-encoding.

Either the order of the check for 6-byte and 3-byte encoding has to be changed, or an easy fix woud be to change the line

if( (x & 0xF0) == 0xE0 && (y & 0xC0) == 0x80 && (z & 0xC0) == 0x80) {

to

if( (x & 0xF0) == 0xE0 && (y & 0xC0) == 0x80 && (z & 0xC0) == 0x80    
     && (x != 0xED || (y & 0xF0) != 0xA0) ) {

so the 3-byte -special case will continue to the if 6-byte check.