Unicode defines more line terminators than what is tested; besides the usual LF, CR and CRLF there are also VT, FF, NEL, LS and PS, but the kata only tests for NEL and LS besides the first three.

There are no random tests.

It's called information :D
You can always on the same level of the solution interchange memory and runtime

IMO, production coding is not hack-competition like CodeWars.
Therefore such signature refactoring is actually welcome and recommended if appropriate.

you are right

Please note that all character offsets are in UTF-8 Code Units, not Grapheme Cluster offset as specified in code comment.

Are you sure that the test 3 and 4 are ok?
Just looks like there are symbols that not equals "\n" like "
" and "…".
And in some parts try to get symbol out of rage like in "AB가
말
語
C"
the length is 10 but try to get the 17th

Fixed!

The test assertions should use XCTAssertEqual(actual, expected) instead of XCTAssertTrue(bool) as the latter does not give any meaningful information why tests failed.

• No random tests for linechar or char
• Accessing elements by indices is a worthless kata idea

Fixed!

I'm working on it. Thanks.

Issues:

1. No random tests
2. Incorrect initial solution (twice char instead of linechar, {})

Btw. This is not the way how Python code should be documented (look at PEP 257)

Is this something called compile-time execution? :P