Question

I have to find a few examples that show that the built in Haskell function “unlines” is not the exact reverse of “lines”, i.e,

unlines.lines x != x

where x is a String containing newlines. I found one example for that:

"aa\nbb"

will become

"aa\nbb\n"

Does anyone know any other examples, that don’t show the same flaw (i.e., unlines always appends a newline after the last line)?

To clearify: According to the automated tests, I already solved the assignment – I simply used the above three times. I’m just interested if there is any other, fundamentally different solution.

Answer 1

If you use quickcheck you can find a smaller one:

import Test.QuickCheck
main = quickCheck $ \x -> (unlines . lines) x == x

It will automatically shrink counterexamples, so it will typically only report the smallest one it finds which is "a".

We can change the property to test only strings ending with \n:

import Test.QuickCheck
main = quickCheck $ \x0 -> let x = x0 ++ "\n" in (unlines . lines) x == x

I did not find any counterexamples.

You might also be interested in testing the other direction, (lines . unlines). Excluding cases where one of the strings contain a newline, this property holds.