Search
Watch video

Video tutorial

NonEmptyList

NonEmptyList is a data type used in Λrrow to model ordered lists that have at least one value. NonEmptyList is available in the arrow-core-data module under the import arrow.core.NonEmptyList

// gradle
compile "io.arrow-kt:arrow-core-data:$arrow_version"


// namespace
import arrow.core.*

of

A NonEmptyList guarantees the list always has at least 1 element.


NonEmptyList.of(1, 2, 3, 4, 5) // NonEmptyList<Int>
NonEmptyList.of(1, 2) // NonEmptyList<Int>
//NonEmptyList.of() // does not compile

Unlike List#[0], NonEmptyList#head is a safe operation that guarantees no exception throwing.

NonEmptyList.of(1, 2, 3, 4, 5).head

foldLeft

When we fold over a NonEmptyList, we turn a NonEmptyList< A > into B by providing a seed value and a function that carries the state on each iteration over the elements of the list. The first argument is a function that addresses the seed value, this can be any object of any type which will then become the resulting typed value. The second argument is a function that takes the current state and element in the iteration and returns the new state after transformations have been applied.


fun sumNel(nel: NonEmptyList<Int>): Int =
  nel.foldLeft(0) { acc, n -> acc + n }

sumNel(NonEmptyList.of(1, 1, 1, 1))
// 4

map

map allows us to transform A into B in NonEmptyList< A >


NonEmptyList.of(1, 1, 1, 1).map { it + 1 }
// NonEmptyList(all=[2, 2, 2, 2])

flatMap

flatMap allows us to compute over the contents of multiple NonEmptyList< * > values


val nelOne: NonEmptyList<Int> = NonEmptyList.of(1)
val nelTwo: NonEmptyList<Int> = NonEmptyList.of(2)

nelOne.flatMap { one ->
  nelTwo.map { two ->
    one + two
  }
}
// NonEmptyList(all=[3])

Monad binding

Λrrow allows imperative style comprehensions to make computing over NonEmptyList values easy.


import arrow.typeclasses.*
import arrow.core.extensions.*

val nelOne: NonEmptyList<Int> = NonEmptyList.of(1)
val nelTwo: NonEmptyList<Int> = NonEmptyList.of(2)
val nelThree: NonEmptyList<Int> = NonEmptyList.of(3)

NonEmptyList.fx {
  val (one) = nelOne
  val (two) = nelTwo
  val (three) = nelThree
  one + two + three
}
// NonEmptyList(all=[6])

Monad binding in NonEmptyList and other collection related data type can be used as generators


NonEmptyList.fx {
  val (x) = NonEmptyList.of(1, 2, 3)
  val (y) = NonEmptyList.of(1, 2, 3)
  x + y
}
// NonEmptyList(all=[2, 3, 4, 3, 4, 5, 4, 5, 6])

Applicative Builder

Λrrow contains methods that allow you to preserve type information when computing over different NonEmptyList typed values.


import arrow.core.*
import java.util.*
import arrow.core.extensions.nonemptylist.apply.map

data class Person(val id: UUID, val name: String, val year: Int)

// Note each NonEmptyList is of a different type
val nelId: NonEmptyList<UUID> = NonEmptyList.of(UUID.randomUUID(), UUID.randomUUID())
val nelName: NonEmptyList<String> = NonEmptyList.of("William Alvin Howard", "Haskell Curry")
val nelYear: NonEmptyList<Int> = NonEmptyList.of(1926, 1900)

map(nelId, nelName, nelYear) { (id, name, year) ->
  Person(id, name, year)
}
// NonEmptyList(all=[Person(id=fb19b7f1-80ed-4540-a5e4-4491c1df4d15, name=William Alvin Howard, year=1926), Person(id=fb19b7f1-80ed-4540-a5e4-4491c1df4d15, name=Haskell Curry, year=1926), Person(id=60f4536e-9204-422c-9ff1-1358320136b7, name=William Alvin Howard, year=1926), Person(id=60f4536e-9204-422c-9ff1-1358320136b7, name=Haskell Curry, year=1926), Person(id=fb19b7f1-80ed-4540-a5e4-4491c1df4d15, name=William Alvin Howard, year=1900), Person(id=fb19b7f1-80ed-4540-a5e4-4491c1df4d15, name=Haskell Curry, year=1900), Person(id=60f4536e-9204-422c-9ff1-1358320136b7, name=William Alvin Howard, year=1900), Person(id=60f4536e-9204-422c-9ff1-1358320136b7, name=Haskell Curry, year=1900)])

Summary

  • NonEmptyList is used to model lists that guarantee at least one element
  • We can easily construct values of NonEmptyList with NonEmptyList.of
  • foldLeft, map, flatMap, and others are used to compute over the internal contents of a NonEmptyList value.
  • fx { ... } comprehensions can be used to imperatively compute over multiple NonEmptyList values in sequence.
  • NonEmptyList.applicative().map { ... } can be used to compute over multiple NonEmptyList values preserving type information and abstracting over arity with map