A pipeline-friendly library for building JSON decoders.

elm-package install NoRedInk/elm-decode-pipeline 3.0.0



A library for building decoders using the pipeline (|>) operator and plain function calls.


It's common to decode into a record that has a type alias. Here's an example of this from the map3 docs:

type alias Job = { name : String, id : Int, completed : Bool }

point : Decoder Job
point =
  map3 Job
    (field "name" string)
    (field "id" int)
    (field "completed" bool)

This works because a record type alias can be called as a normal function. In that case it accepts one argument for each field (in whatever order the fields are declared in the type alias) and then returns an appropriate record built with those arguments.

The mapN decoders are straightforward, but require manually changing N whenever the field count changes. This library provides functions designed to be used with the |> operator, with the goal of having decoders that are both easy to read and easy to modify.


Here is a decoder built with this library.

import Json.Decode exposing (int, string, float, Decoder)
import Json.Decode.Pipeline exposing (decode, required, optional, hardcoded)

type alias User =
  { id : Int
  , email : Maybe String
  , name : String
  , percentExcited : Float

userDecoder : Decoder User
userDecoder =
  decode User
    |> required "id" int
    |> required "email" (nullable string) -- `null` decodes to `Nothing`
    |> optional "name" string "(fallback if name is `null` or not present)"
    |> hardcoded 1.0

In this example:

  • decode is a synonym for succeed (it just reads better here)
  • required "id" int is similar to ("id" := int)
  • optional is like required, but if the field is either null or not present, decoding does not fail; instead it succeeds with the provided fallback value.
  • hardcoded does not look at the provided JSON, and instead always decodes to the same value.

You could use this decoder as follows:

    {"id": 123, "email": "", "name": "Sam Sample"}

The result would be:

{ id = 123
, email = ""
, name = "Sam Sample"
, percentExcited = 1.0

Alternatively, you could use it like so:

    {"id": 123, "email": "", "percentExcited": "(hardcoded)"}

In this case, the result would be:

{ id = 123
, email = ""
, name = "(fallback if name not present)"
, percentExcited = 1.0