So there are hidden constraints other than these?

• must be equal to the number of lines in the code)
• and no spacing characters (spaces, tabs...)

I love code golf katas, and my impression has been that they generally tend to be underrated for whatever reason.

(And I'm also stuck, 217 characters and counting...)

First of all I would like to thank you for this kata. It was a very interesting and good challenge. First of all, I am sorry that after 4 years it is still in beta, that few people are trying to solve it and only a few have solved it. I would really like to see it get some attention and get it to release.
I would like to add that some corner cases are not tested.

1. For example, there are only three maps in fixed tests where there is a first aid kit. Perhaps we should add such a case to the random tests?
2. Also not tested are maps in which enemies are near the player's starting position on the map. That is, cases when the player in any case initially receives damage, for example, something like this. Perhaps we should take into account such cases?

field = [
  'SHM   |EX',
  '###   ###',
  'CCK     C',
  ' E      E',
  'DEE    E<',
]

3. All the cards except the one you gave as an example in the description are 5x9. I think we should do some tests with larger maps?

These are just thoughts after solving the problem, perhaps you have a different opinion on this? Anyway, thank you very much, the evening was well spent:D

Shouldn't this maze return True? If you look at this path and count steps (blue dots), then they sum exactly equal to 10, as in requirement (print at the bottom). And health (3-1+2-3) is ok. Am i missing something?

https://www.codewars.com/kata/reviews/5e9a335aa66a4c00019dc931/groups/5e9c6a7536c5250001495cb4

The description has been out of sync from the tests; the example still writes e_digit.

I changed the sequential tests to the same number of repetitions as the random ones. That many repetitions were unnecessary anyways :/

It certainly makes much more sense this way. (Though really you should put the computations independent of text inside the constructor in your reference solution)

I was a little bit skeptic of this in the beggining, but I think it might be benefitial for some more creative answers, so I removed the variable-length example texts and added a common AffineCipher object that is reused many times for each source. Also updated the description.

Do you think it's better this way?

Oops, done!

Title needs to be updated too.

Woah, I didn't know that. I updated the description.

Thank you :)