Now I remember this puzzle. It is one of the very few (the only one?) puzzles I need to enable UTF-8 encoding in my source to make it pass in CG.
Changing the test case will fail all existing solutions. It seems it is not a good idea to do it this way.
I am thinking if I were to write the solution in C, it could have been a painful experience. I wrote the code in a local editor. Directly writing the code in CG’s online editor may have another (easier?) experience I haven’t tested it out.
I think we can just let it be. If kinds of improvement is really necessary, it could be a “warning” in the Input section notifying coders they have to pay attention to the non-ASCII character in input.