Skip to content
Type mixing
Branch: master
Clone or download

Alphabet soup

This library is intended to give a seamless way to manipulate scala structures into one another, mixing the types intelligently as required

For a whistle-stop tour of the underlying algorithm, see here


Releases are available on Maven. Add the following to your build.sbt:

"io.typechecked" %% "alphabet-soup" % "<version>"

for the version of your choice. The project is fully tagged and release versions are available to view online.



The top-level concept is called Mixer. You give it two types, and if the libary can squeeze the first into the second then it compiles.

At runtime, it turns the supplied value into an instance of the second value.

A quick example:

case class Pet(name: String, age: Int)

val pet: Pet = Pet("Jeremy", 5)

Mixer[Pet, (Int, String)].mix(pet) == (5, "Jeremy")

It works recursively on much larger examples too.


The base element in the library is an Atom. Mixer stops searching the branch of the tree it's in once it reaches an Atom - so they form the building blocks of the structures you want to manipulate.

The library provides Atom evidence for the usual suspects Int, String, etc.

Making your own is very simple:

implicit val atomT: Atom[T] = Atom[T]

Just make sure it's in scope whenever you create a Mixer.

Alternatively an Atom can be created by using the macro annotation @Atomic.

In order to use the macro you must add the following compiler plugin to your build.sbt

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full),

Then creating an Atom is simple:

@Atomic case class Foo(a: Int)
@Atomic trait Foo
@Atomic class Foo


A Molecule is something beyond a boundary that we can't handle at compile time - for example a List. The information of how many items are in the source List is not known to us, so we can't do anything without some extra help or structure.

That takes the form of recursing inside the Molecule, and treating it as its own closed off world of Atoms and sub-Molecules to process. No value from an outer class can make it into a Molecule.

A Molecule is mixed according to the normal rules:

trait A
trait B
case class AS(as: List[A])
case class BS(bs: List[B])

In the above example, AS can be mixed into BS if and only if A can be mixed into B.

List is provided for you in the library, for example. In general any Functor can be implemented very simply as a molecule.


There are two ways of creating defaults in alphabet-soup.

The first is to create a DefaultAtom[T]:

case class A(i: Int)
case class B(i: Int, s: String)

implicit val default: DefaultAtom[String] = DefaultAtom("some default string")

val mixer = Mixer[A,B]


This produces B(1, "some default string")

This can be useful if you want to create default values for types which do not possess much information.

HNil and Unit have provided DefaultAtoms.

The second way of creating defaults often comes in handy when you want to mix a class into a bigger type on a one-time basis. To do this, you must manually create your Mixer using the builder provided, and give it the required defaults. For example:

case class A(i: Int)
case class B(i: Int, s: String, b: Boolean)

Mixer[A, B]  // Does not compile
val m = Mixer.from[A].to[B].withDefault("").withDefault(true).build  // Does compile. This is an instance of `Mixer[A, B]`
m.mix(A(0)) == B(0, "", true)

Complex example

Here's a more complicated example to show what it can do.

// You should follow the philosophy of every value is a type!
@Atomic case class FirstName(value: String)
@Atomic case class LastName(value: String)
@Atomic case class Address1(value: String)
@Atomic case class City(value: String)
@Atomic case class Postcode(value: String)
@Atomic case class Title(value: String)
@Atomic case class Gender(value: String)

// This is our data tree
case class Address(a1: Address1, c: City, p: Postcode)
case class Alias(firstName: FirstName, lastName: LastName, isLegal: Boolean)
case class AddressHistory(values: List[Address])

// This is our source data class
case class Source(
  firstName: FirstName,
  lastName: LastName,
  addressHistory: AddressHistory,
  aliases: List[(Title, FirstName, LastName)]

// This is what we'll be mapping it into
case class Target(
  name: (FirstName, LastName),
  addresses: List[(Address1, Postcode)],
  aliases: List[Alias],
  gender: Gender

// We're going to map an intance of Source into an instance of Target
val source = Source(
  firstName = FirstName("John"),
  lastName = LastName("Johnson"),
  addressHistory = AddressHistory(List(
    Address(Address1("5 John Street"), City("Johnsville"), Postcode("JOHN")),
    Address(Address1("5 Jack Street"), City("Jacksville"), Postcode("JACK"))
  aliases = List(
    (Title("Mr"), FirstName("Johnny"), LastName("Vegas")),
    (Title("Mr"), FirstName("Jon"), LastName("Snow"))

// We need a Mixer

// First attempt:
Mixer[Source, Target]  // Won't compile because we have no 'Gender' in our Source

// Second attempt:
Mixer.from[Source].to[Target].withDefault(Gender("male")).build  // Won't compile because we have no 'isLegal' in our source aliases

// Third attempt:
implicit val submixer = Mixer.from[(Title, FirstName, LastName)].to[Alias].withDefault(true)
val mixer = Mixer.from[Source].to[Target].withDefault(Gender("male")).build

val result = mixer.mix(source)

// This test passes!
result shouldBe Target(
  name = (FirstName("John"), LastName("Johnson")),
  addresses = List(
    (Address1("5 John Street"), Postcode("JOHN")),
    (Address1("5 Jack Street"), Postcode("JACK"))
  aliases = List(
    Alias(FirstName("Johnny"), LastName("Vegas"), true),
    Alias(FirstName("Johnny"), LastName("Vegas"), true)
  gender = Gender("male")

The future

This library will be extended to:

  • handle functional Atoms, which contain within them a way to transform one type to another
  • have a "strict" version, which will prevent duplicate types in the source of the mixer
You can’t perform that action at this time.