davegurnell / bridges   0.24.0

GitHub

Generate bindings for Scala types in other programming languages.

Scala versions: 2.13 2.12

Bridges

Generate bindings for Scala types in other programming languages.

Copyright 2017-2019 Dave Gurnell. Licensed Apache 2.0.

Maintainers:

  • Dave Gurnell (Flow and Typescript support)
  • Pere Villega (Elm support)

Build Status Coverage status Maven Central

Getting Started

Grab the code by adding the following to your build.sbt:

libraryDependencies += "com.davegurnell" %% "bridges" % "<<VERSION>>"

Synopsis

Render Typescript/Flow Declarations for Scala ADTs

First, create a simple data model:

final case class Color(red: Int, green: Int, blue: Int)

sealed abstract class Shape extends Product with Serializable
final case class Circle(radius: Double, color: Color) extends Shape
final case class Rectangle(width: Double, height: Double, color: Color) extends Shape

The bridges.typescript and bridges.flow packages expose similar APIs for encoding ADTs as structural types.

The decl[Foo] method creates an object representing a type declaration for Foo.

The render(...) method writes a list of declaration objects to a String in the relevant target language. Here's an example for Typescript:

import bridges.typescript._
import bridges.typescript.syntax._

Typescript.render(List(
  decl[Color],
  decl[Circle],
  decl[Rectangle],
  decl[Shape]
))
// res1: String =
// export interface Color {
//   red: number;
//   green: number;
//   blue: number;
// };
//
// export interface Circle {
//   radius: number;
//   color: Colo;
// };
//
// export interface Rectangle {
//   width: number;
//   height: number;
//   color: Color;
// };
//
// export type Shape =
//   {
//     type: "Circle",
//     radius: number,
//     color: Color
//   } |
//   {
//     type: "Rectangle",
//     width: number,
//     height: number,
//     color: Color
//   };

Simply replace the imports to target Flow instead:

import bridges.flow._
import bridges.flow.syntax._

Flow.render(List(
  decl[Color],
  decl[Circle],
  decl[Rectangle],
  decl[Shape]
))

The syntax packages also provide a simple DSL for defining structural types directly:

import bridges.typescript._
import bridges.typescript.TsType._
import bridges.typescript.syntax._

val logMessage: TsDecl =
  decl("LogMessage")(struct(
    "level" --> union(lit("error"), lit("warning")),
    text    --> Str
  )

Typescript.render(logMessage)
// res0: String =
// export interface LogMessage {
//   level: "error" | "warning";
//   text: string;
// };

You can even create generic types using the DSL, which is something the shapeless derivation currently can't handle:

import bridges.typescript._
import bridges.typescript.TsType._
import bridges.typescript.syntax._

val pair: TsDecl =
  decl("Pair", "A", "B")(struct(
    "head" --> Ref("A"),
    "tail" --> Ref("B"),
  )

Typescript.render(pair)
// res0: String =
// export interface Pair<A, B> {
//   head: A;
//   tail: B;
// };

Render Elm Definitions and JSON Codecs for Scala ADTs

The bridges.elm package generates type definitions and JSON codecs for Elm:

  • Elm.render generates a type definition for a list of declarations;
  • Elm.jsonDecoder generates an Elm Decode.Decoder
  • Elm.jsonEncoder generates an Elm Encoder
  • Elm.buildFile returns a pair (String, String) where the first element is the name of an Elm file and the second element is the content for that file, containing type and JSON codec definitions.

To avoid circular references, Elm.buildFile can generate a single file for a list of declarations.

If you want to use any of the JSON encoders or decoders generated by the project you need to do the following:

  • Your Elm project must include the NoRedInk/elm-decode-pipeline dependency
  • We assume any non-primitive type in your ADT will be generated by Bridges in the same module as the current ADT, to be able to define the right imports for them.
  • If your ADT contains a sum type, the generated json must distinguish between alternatives using a field named type that encodes the name of the product instance as a String. If you use Circe, see this link.

NOTE: automatic encoder and decoder generation doesn't work for types with Generics. You will need to manually create your own.

Working with Refined types

If you are interested in this library you are most likely using Refined.

We have provided a default encoder for refined types. It will defaults to the basic type associated with the refined type. For example:

  • For Int Refined Greater[W.6.T] we treat the type as an Int
  • For String Refined Size[ClosedOpen[W.1.T, W.100.T]] we treat the type as a String
  • etc

This should cover most (if not all) use cases of refined types when converting to other languages. You can still override the default encoder with your own higher-precedence encoder.

You can see an example of this in tests for class ClassWithRefinedType.

Developing

  • Development should be completed on feature branches and submitted by PR to master.
  • Commits to master are published as snapshot builds.
  • Tags of the form x.y.z are published as release builds.