Watch video

Video tutorial

Either

beginner

In day-to-day programming, it is fairly common to find ourselves writing functions that can fail. For instance, querying a service may result in a connection issue, or some unexpected JSON response.

To communicate these errors it has become common practice to throw exceptions; however, exceptions are not tracked in any way, shape, or form by the compiler. To see what kind of exceptions (if any) a function may throw, we have to dig through the source code. Then to handle these exceptions, we have to make sure we catch them at the call site. This all becomes even more unwieldy when we try to compose exception-throwing procedures.

import arrow.*
import arrow.core.*

val throwsSomeStuff: (Int) -> Double = {x -> x.toDouble()}
val throwsOtherThings: (Double) -> String = {x -> x.toString()}
val moreThrowing: (String) -> List<String> = {x -> listOf(x)}
val magic = throwsSomeStuff.andThen(throwsOtherThings).andThen(moreThrowing)
magic
// (A) -> C

Assume we happily throw exceptions in our code. Looking at the types of the above functions, any of them could throw any number of exceptions – we do not know. When we compose, exceptions from any of the constituent functions can be thrown. Moreover, they may throw the same kind of exception (e.g. IllegalArgumentException) and thus it gets tricky tracking exactly where an exception came from.

How then do we communicate an error? By making it explicit in the data type we return.

Either vs Validated

In general, Validated is used to accumulate errors, while Either is used to short-circuit a computation upon the first error. For more information, see the Validated vs Either section of the Validated documentation.

By convention the right hand side of an Either is used to hold successful values.

val right: Either<String, Int> = Either.Right(5)
right
// Right(b=5)
val left: Either<String, Int> = Either.Left("Something went wrong")
left
// Left(a=Something went wrong)

Because Either is right-biased, it is possible to define a Monad instance for it.

Since we only ever want the computation to continue in the case of Right (as captured by the right-bias nature), we fix the left type parameter and leave the right one free.

So the map and flatMap methods are right-biased:

val right: Either<String, Int> = Either.Right(5)
right.flatMap{Either.Right(it + 1)}
// Right(b=6)
val left: Either<String, Int> = Either.Left("Something went wrong")
left.flatMap{Either.Right(it + 1)}
// Left(a=Something went wrong)

Using Either instead of exceptions

As a running example, we will have a series of functions that will:

  • Parse a string into an integer
  • Calculate the reciprocal
  • Convert the reciprocal into a string

Using exception-throwing code, we could write something like this:

// Exception Style

fun parse(s: String): Int =
    if (s.matches(Regex("-?[0-9]+"))) s.toInt()
    else throw NumberFormatException("$s is not a valid integer.")

fun reciprocal(i: Int): Double =
    if (i == 0) throw IllegalArgumentException("Cannot take reciprocal of 0.")
    else 1.0 / i

fun stringify(d: Double): String = d.toString()

Instead, let’s make the fact that some of our functions can fail explicit in the return type.

// Either Style

fun parse(s: String): Either<NumberFormatException, Int> =
    if (s.matches(Regex("-?[0-9]+"))) Either.Right(s.toInt())
    else Either.Left(NumberFormatException("$s is not a valid integer."))

fun reciprocal(i: Int): Either<IllegalArgumentException, Double> =
    if (i == 0) Either.Left(IllegalArgumentException("Cannot take reciprocal of 0."))
    else Either.Right(1.0 / i)

fun stringify(d: Double): String = d.toString()

fun magic(s: String): Either<Exception, String> =
    parse(s).flatMap{reciprocal(it)}.map{stringify(it)}

These calls to parse returns a Left and Right value

parse("Not a number")
// Left(a=java.lang.NumberFormatException: Not a number is not a valid integer.)
parse("2")
// Right(b=2)

Now, using combinators like flatMap and map, we can compose our functions together.

magic("0")
// Left(a=java.lang.IllegalArgumentException: Cannot take reciprocal of 0.)
magic("1")
// Right(b=1.0)
magic("Not a number")
// Left(a=java.lang.NumberFormatException: Not a number is not a valid integer.)

In the following exercise we pattern-match on every case the Either returned by magic can be in. Note the when clause in the Left - the compiler will complain if we leave that out because it knows that given the type Either[Exception, String], there can be inhabitants of Left that are not NumberFormatException or IllegalArgumentException. You should also notice that we are using SmartCast for accessing to Left and Right value.

val x = magic("2")
val value = when(x) {
    is Either.Left -> when (x.a){
        is NumberFormatException -> "Not a number!"
        is IllegalArgumentException -> "Can't take reciprocal of 0!"
        else -> "Unknown error"
    }
    is Either.Right -> "Got reciprocal: ${x.b}"
}
value
// Got reciprocal: 0.5

Instead of using exceptions as our error value, let’s instead enumerate explicitly the things that can go wrong in our program.

// Either with ADT Style

sealed class Error {
    object NotANumber : Error()
    object NoZeroReciprocal : Error()
}

fun parse(s: String): Either<Error, Int> =
        if (s.matches(Regex("-?[0-9]+"))) Either.Right(s.toInt())
        else Either.Left(Error.NotANumber)

fun reciprocal(i: Int): Either<Error, Double> =
        if (i == 0) Either.Left(Error.NoZeroReciprocal)
        else Either.Right(1.0 / i)

fun stringify(d: Double): String = d.toString()

fun magic(s: String): Either<Error, String> =
        parse(s).flatMap{reciprocal(it)}.map{stringify(it)}

For our little module, we enumerate any and all errors that can occur. Then, instead of using exception classes as error values, we use one of the enumerated cases. Now when we pattern match, we are able to comphrensively handle failure without resulting to an else branch; moreover since Error is sealed, no outside code can add additional subtypes which we might fail to handle.

val x = magic("2")
when(x) {
    is Either.Left -> when (x.a){
        is Error.NotANumber -> "Not a number!"
        is Error.NoZeroReciprocal -> "Can't take reciprocal of 0!"
    }
    is Either.Right -> "Got reciprocal: ${x.b}"
}

Syntax

Either can also map over the left value with mapLeft which is similar to map but applies on left instances.

val r : Either<Int, Int> = Either.Right(7)
r.mapLeft {it + 1}
val l: Either<Int, Int> = Either.Left(7)
l.mapLeft {it + 1}
// Left(a=8)

Either<A, B> can be transformed to Either<B,A> using the swap() method.

val r: Either<String, Int> = Either.Right(7)
r.swap()
// Left(a=7)

For using Either’s syntax on arbitrary data types. This will make possible to use the left(), right(), contains(), getOrElse() and getOrHandle() methods:

7.right()
// Right(b=7)
"hello".left()
// Left(a=hello)
val x = 7.right()
x.contains(7)
// true
val x = "hello".left()
x.getOrElse { 7 }
// 7
val x = "hello".left()
x.getOrHandle { "$it world!" }
// hello world!

For creating Either instance based on a predicate, use Either.cond() method :

Either.cond(true, { 42 }, { "Error" })
// Right(b=42)
Either.cond(false, { 42 }, { "Error" })
// Left(a=Error)

Another operation is fold. This operation will extract the value from the Either, or provide a default if the value is Left

val x : Either<Int, Int> = 7.right()
x.fold({ 1 }, { it + 3 }) 
// 10
val y : Either<Int, Int> = 7.left()
y.fold({ 1 }, { it + 3 }) 
// 1

The getOrHandle() operation allows the transformation of an Either.Left value to a Either.Right using the value of Left. This can be useful when a mapping to a single result type is required like fold() but without the need to handle Either.Right case.

As an example we want to map an Either<Throwable, Int> to a proper HTTP status code:

val r: Either<Throwable, Int> = Either.Left(NumberFormatException())
val httpStatusCode = r.getOrHandle {
	when(it) {
		is NumberFormatException -> 400
		else -> 500
	}
} // 400

The leftIfNull operation transforms a null Either.Right value to the specified Either.Left value. If the value is non-null, the value wrapped into a non-nullable Either.Right is returned (very useful to skip null-check further down the call chain). If the operation is called on an Either.Left, the same Either.Left is returned.

See the examples below:

Right(12).leftIfNull({ -1 })
// Right(b=12)
Right(null).leftIfNull({ -1 })
// Left(a=-1)
 Left(12).leftIfNull({ -1 })
// Left(a=12)

Another useful operation when working with null is rightIfNotNull. If the value is null it will be transformed to the specified Either.Left and if its not null the type will be wrapped to Either.Right.

Example:

"value".rightIfNotNull { "left" }
// Right(b=value)
null.rightIfNotNull { "left" }
// Left(a=left)

Arrow contains Either instances for many useful typeclasses that allows you to use and transform right values. Both Option and Try don’t require a type parameter with the following functions, but it is specifically used for Either.Left

Functor

Transforming the inner contents

import arrow.instances.either.functor.*
 
Right(1).map {it + 1}
// Right(b=2)

Applicative

Computing over independent values

import arrow.instances.either.applicative.*
  
tupled(Either.Right(1), Either.Right("a"), Either.Right(2.0))
// Right(b=Tuple3(a=1, b=a, c=2.0))

Monad

Computing over dependent values ignoring absence

import arrow.instances.either.monad.*

binding {
  val a = Either.Right(1).bind()
  val b = Either.Right(1 + a).bind()
  val c = Either.Right(1 + b).bind()
  a + b + c
}
// Right(6)

Supported type classes

Module Type classes
arrow.aql Count, From, GroupBy, OrderBy, Select, Sum, Union
arrow.optics.typeclasses Each
arrow.typeclasses ApplicativeError, Applicative, Bifunctor, Eq, Foldable, Functor, Hash, MonadError, Monad, Monoid, Semigroup, SemigroupK, Show, Traverse