avh4-experimental/elm-transducers

Composable transformation of sequences using clojure-inspired transducers

Source code at GitHub https://package.elm-lang.org/packages/avh4-experimental/elm-transducers/latest/

{
  "createdAt": "2015-06-03T02:48:00Z",
  "defaultBranch": "master",
  "description": "Composable transformation of sequences using clojure-inspired transducers",
  "fullName": "avh4-experimental/elm-transducers",
  "homepage": "https://package.elm-lang.org/packages/avh4-experimental/elm-transducers/latest/",
  "language": "Elm",
  "name": "elm-transducers",
  "pushedAt": "2018-08-31T07:45:45Z",
  "stargazersCount": 40,
  "topics": [],
  "updatedAt": "2025-02-06T13:54:39Z",
  "url": "https://github.com/avh4-experimental/elm-transducers"
}

elm-transducers

A transducer is a composable way of processing a series of values. Many basic transducers correspond to functions you may be familiar with for processing Lists.

Transducers enable efficient processing

Transducers can be used to combine processing operations in a way that allows processing to be done more efficiently.

When using List.map, it is more efficient to compose multiple functions and then map the list with the composed function than to map the list with each function independently because the list will only be traversed once. Similarly, transducers can be used to process Lists more efficiently, but it is not limited to mapping operations. filter, take, drop, and any other transducer can be efficiently composed.

import List as L
import Transducer as T exposing ((>>>))

slowMapChain = [1, 2, 3] |> L.map ((+) 10) |> L.map toString
fastMapChain = [1, 2, 3] |> L.map ((+) 10 >> toString)

slowChain = [1, 2, 3] |> L.filter ((/=) 2) |> L.map toString
fastChain = [1, 2, 3] |> T.transduceList (T.filter ((/=) 2) >>> T.map toString)

Transducers can be reused

Transducers can be reused with many different data types. List, Array, Set, Dict are supported by this library, and you can define your own transducer processes to work with other data types. You can also define transducer processes that convert between types (for example, transducing from a List into a Set).

import Maybe
import String
import Transducer as T exposing ((>>>))
import Result exposing (toMaybe)
import Set exposing (Set)


parseValidInts =
    T.map String.toInt
        >>> T.map toMaybe
        >>> T.filter ((/=) Nothing)
        >>> T.map (Maybe.withDefault 0)


exampleList : List Int
exampleList =
    T.transduceList parseValidInts [ "123", "-34", "35.0", "SDF", "7" ]


exampleConvert : Set Int
exampleConvert =
    T.transduce List.foldr Set.insert Set.empty parseValidInts [ "123", "-34", "35.0", "SDF", "7" ]

Differences from clojure’s tranducers

In elm, it is more natural for the Reducer type to be a -> r -> r instead of r -> a -> r
Elm is pure, meaning that elm transducers cannot hide state. As a result, transducers cannot simply be functions as they are in clojure. This also means that the transducer type must include a type parameter for the transducer’s state.

avh4-experimental/elm-transducers

elm-transducers

Transducers enable efficient processing

Transducers can be reused

Differences from clojure’s tranducers

See also