Either

Either is one of those types that you already know in Swift but it’s specialized in some values.

Here is it:

enum Either<Left, Right> {
    case left(Left)
    case right(Right)
}

Slo it’s an enum with two cases. That should ring at least two bells. Usually the right one is associated with the good or success or dextra. The left on the other hand is for bad, failure, error or you can say sinistra.

At least two types that are common in Swift are a specialization of this Either.

You can take a look at Optional and match left with none case. Also you can take a look at Result and match if failure with a left case.

So here you have a more generic type that you can use. It’s useful not only when you have to validate stuff. But also when there is a need to return different types.

Computed Properties

To make life a bit easier there are defined some helper properties.

Handy properties for logic checks.

var isLeft : Bool // true if `left` case
var isRight: Bool // true if `right` case

When you want to get to a value but you can deal with optional.

var right: Right?
var left : Left?

Take a look at OptionalAPI to see how working with Optional can be a pleasure.

maps

func map<R>(_ transform: (Right) -> R) -> Either<Left,R>

It’s a function that takes a function expecting an instance of Right and produces a new instance of Either<Left, R>. Type R means NewRight but is abbreviated.

Either treats it’s right value a bit specially. This is analogous how Result and Optional treat some of their own cases.

That means that if you want to map an Either then by default you will be given an instance of a Right type.

So if you have:

func increment(_ i: Int) -> Int { i + 1 }

let right = Either<String,Int>.right(42)

right
    .map(increment) // .right(43)

Final result is a new instance of Either<String,Int> with it’s right case holding 43. Same thing with left would do nothing.

let left = Either<String,Int>.left("I'm left")

left
    .map(increment) //  .left("I'm left")

rightMap

func mapRight<R>(_ transform: (Right) -> R) -> Either<Left, R>

You can be explicit about it and call rightMap that is just a wrapper around map.

leftMap

func mapLeft<L>(_ transform: (Left) -> L) -> Either<L, Right>

However if you want to transform a left value you can use leftMap. Same story as for right but for left.

left
    .map({ $0.uppercased() }) // .left("I'M LEFT")

biMap

func biMap<L,R>( _  leftTransform:  (Left) -> L,
                 _ rightTransform: (Right) -> R)
     -> Either<L, R>

This function takes two other functions as arguments. Left is the left transform and the right is the right transform. This results in a new instance of Either<L,R>.

This one is a map that combines leftMap and rightMap. More common name for it is biMap but I would like it be handy in the IDE and you can start typing just map and see what’s there.

This biMap can be used when you want to combine those transformations in to one statement:

right
    .biMap({ $0.uppercased() }, increment) // .right(43)

left
    .biMap({ $0.uppercased() }, increment) // .left("I'M LEFT")

flatMaps

Same as map above, but with one difference.

func flatMap<R>(
    _ transform: (Right) -> Either<Left, R> 
    ) -> Either<Left, R>

func flatMapLeft<L>(
    _ transform: (Left) -> Either<L, Right>
    ) -> Either<L, Right>

This time a transform function returns another Either. This wold result in an Either Either like Either<Left, Either<Left, R>>. So using this flatMap (or another name you can fins bind) you can remove one layer of nesting.

Just as a side note, whenever you are using flatMap you are doing the dreaded monadic computation. You were calling flatMap on optionals and it was fine. And yes Optional is a Monad. If you want you can Google it just don’t overthink it and you will be just fine 😎

Utils

either

func either<L,R,T>(
    _  leftTransform: @escaping (L) -> T,
    _ rightTransform: @escaping (R) -> T) -> (Either<L,R>) -> T

This might not be so common, but it’s a function that returns another function. The way it works you start with providing two functions. Left transform knows kow to produce T given an L. Right transform knows kow to produce T given an R.

Next step is the returned function. This one expects and Either<L,R> and when you provide an instance of this either then it will produce and instance of T.

lefts

func lefts<L,R>(_ eithers: [Either<L,R>] ) -> [L]

I guess type says it all. Give this function an array of either-s and in return you will get an array of Ls.

rights

func rights<L,R>(_ eithers: [Either<L,R>] ) -> [R]

I guess type says it all. Give this function an array of either-s and in return you will get an array of Rs.

partitionEithers

Returns a tuple containing all L values in the left/first array. And all the R in the right/ second array.

let eithers: [Either<String,Int>] = 
    [ .left("A"), 
      .right(42), 
      .left("B"), 
      .right(24) ]

let (lefts, rights) = partitionEithers(eithers)

lefts  // ["A", "B"]
rights // [42, 24]

YouTube

If you know 🇵🇱 then you can check out this YT playlist Either - Monada Either w Swift

🐇🕳 Rabbit Hole

This project is part of the 🐇🕳 Rabbit Hole Packages Collection