I've had a look at modifying both implementations we have for the normal Array case detailed in #225 for use with a NonEmptyArray. I came up with this one based on my implementation:
import Data.Array.NonEmpty
import Data.Array.NonEmpty.Internal (NonEmptyArray(..))
import Data.Maybe (Maybe(..))
import Prelude (map)
transpose :: forall a. NonEmptyArray (NonEmptyArray a) -> NonEmptyArray (NonEmptyArray a)
transpose x =
case (mapMaybe tail' x) of
[] ->
singleton (map head x)
end ->
cons (map head x) (transpose (unsafeFromArray end))
where
tail' :: NonEmptyArray a -> Maybe (NonEmptyArray a)
tail' x =
case tail x of
[] -> Nothing
end -> Just (unsafeFromArray end)
unsafeFromArray :: forall a. Array a -> NonEmptyArray a
unsafeFromArray = NonEmptyArrayand this one based on the implementation from @JordanMartinez :
import Prelude
import Data.Array as A
import Data.Array.NonEmpty
import Data.Array.NonEmpty.Internal (NonEmptyArray(..))
import Data.Maybe (Maybe(..), maybe)
import Data.Foldable (foldl)
transpose :: forall a. NonEmptyArray (NonEmptyArray a) -> NonEmptyArray (NonEmptyArray a)
transpose xs = unsafeFromArray $ go 0 []
where
go :: Int -> Array (NonEmptyArray a) -> Array (NonEmptyArray a)
go idx allArrays = case buildNext idx of
Nothing -> allArrays
Just next -> go (idx + 1) (A.snoc allArrays next)
buildNext :: Int -> Maybe (NonEmptyArray a)
buildNext idx =
xs # flip foldl Nothing \acc nextArr -> do
maybe acc (\el -> Just $ maybe (singleton el) (flip snoc el) acc) $ index nextArr idx
unsafeFromArray :: forall a. Array a -> NonEmptyArray a
unsafeFromArray = NonEmptyArrayAll the tests pass for whichever version, but performance comparisons show the same pattern as for Array - Jordan's is again about three times faster than mine and is only a little slower than for Array. I'm slightly concerned about the use of unsafeFromArray at the top level, but I think it's legitimate and many other functions in the module rely on it.
If you're agreeable, I'll raise a PR using this.
I've had a look at modifying both implementations we have for the normal
Arraycase detailed in #225 for use with aNonEmptyArray. I came up with this one based on my implementation:and this one based on the implementation from @JordanMartinez :
All the tests pass for whichever version, but performance comparisons show the same pattern as for
Array- Jordan's is again about three times faster than mine and is only a little slower than forArray. I'm slightly concerned about the use ofunsafeFromArrayat the top level, but I think it's legitimate and many other functions in the module rely on it.If you're agreeable, I'll raise a PR using this.