alexarchambault / argonaut-shapeless   1.2.0

BSD 3-clause "New" or "Revised" License GitHub

Automatic argonaut codec derivation with shapeless

Scala versions: 2.13 2.12 2.11 2.10
Scala.js versions: 1.x 0.6
Scala Native versions: 0.4 0.3

argonaut-shapeless

Automatic argonaut codec derivation with shapeless

Build status Maven Central Gitter

It is available for scala 2.12, and 2.13, and depends on argonaut 6.3.

argonaut-shapeless is part of the shapeless ecosystem of Typelevel, and as such endorses the Scala Code of Conduct.

It is one of the very first projects to have used Lazy from shapeless 2.1, which made type class derivation with implicits much more robust.

Compatibility table

argonaut-shapeless argonaut shapeless refined
1.3.0 6.3.3 2.3.x 0.9.x
1.2.0-M4 6.2-RC2 2.3.x 0.6.x
1.2.0-M{1,3} 6.2-M3 2.3.x 0.5.x
1.1.1 6.1a 2.3.x 0.4.x
1.1.0 6.1 2.3.x 0.4.x
1.0.x 6.1 2.2.x 0.3.5
0.3.x 6.1 2.2.x n/a

Usage

Add to your build.sbt

libraryDependencies +=
  "com.github.alexarchambault" %% "argonaut-shapeless_6.3" % "1.3.0"

Features

The examples below assume you imported the content of argonaut, argonaut.Argonaut, and argonaut.ArgonautShapeless, like

import argonaut._, Argonaut._, ArgonautShapeless._

Automatic codecs for case classes

case class CC(i: Int, s: String)

// encoding
val encode = EncodeJson.of[CC]

val json = encode(CC(2, "a"))
json.nospaces == """{"i":2,"s":"a"}"""

// decoding
val decode = DecodeJson.of[CC]

val result = decode.decodeJson(json)
result == DecodeResult.ok(CC(2, "a"))

The way case classes are encoded can be customized, see below.

Automatic codecs for sealed traits

import argonaut._, Argonaut._, ArgonautShapeless._
sealed trait Base
case class First(i: Int) extends Base
case class Second(s: String) extends Base

// encoding
val encode = EncodeJson.of[Base]

val json = encode(First(2))
json.nospaces == """{"First":{"i":2}}"""

// decoding
val decode = DecodeJson.of[Base]

val result = decode.decodeJson(json)
result == DecodeResult.ok(First(2))

Default values

Like upickle, fields equal to their default value are not put in the result JSON object.

case class CC(i: Int = 4, s: String = "foo")

CC().asJson.nospaces == "{}"
CC(i = 3).asJson.nospaces == """{"i":3}"""
CC(i = 4, s = "baz").asJson.nospaces == """{"s":"baz"}"""

"{}".decodeOption[CC] == Some(CC())
"""{"i":2}""".decodeOption[CC] == Some(CC(i = 2))
"""{"s":"a"}""".decodeOption[CC] == Some(CC(s = "a"))

This can be turned off by providing the alwaysIncludeDefaultValue JsonProductCodecFor.

implicit def alwaysIncludeCodecFor[T]: derive.JsonProductCodecFor[T] =
    derive.JsonProductCodecFor.alwaysIncludeDefaultValue

CC().asJson.nospaces == """{"i":4,"s":"foo"}"""

Custom encoding of case classes

When encoding / decoding a case class C, argonaut-shapeless looks for an implicit JsonProductCodecFor[C], which has a field codec: JsonProductCodec. A JsonProductCodec provides a general way of encoding / decoding case classes.

The default JsonProductCodecFor[T] for all types provides JsonProductCodec.obj, which encodes / decodes case classes as shown above.

This default can be changed, e.g. to convert field names to serpent_case,

import argonaut.derive._

implicit def serpentCaseCodecFor[T]: JsonProductCodecFor[T] =
  JsonProductCodecFor(JsonProductCodec.adapt(_.toSerpentCase))

case class Identity(firstName: String, lastName: String)

Identity("Jacques", "Chirac").asJson.nospaces == """{"first_name":"Jacques","last_name":"Chirac"}"""

This can be changed for all types at once like just above, or only for specific types, like

implicit def serpentCaseCodecForIdentity: JsonProductCodecFor[Identity] =
  JsonProductCodecFor(JsonProductCodec.adapt(_.toSerpentCase))

Custom encoding for sealed traits

When encoding / decoding a sealed trait S, argonaut-shapeless looks for an implicit JsonSumCodecFor[S], which has a field codec: JsonSumCodec. A JsonSumCodec provides a general way of encoding / decoding sealed traits.

The default JsonSumCodecFor[S] for all types S provides JsonSumCodec.obj as JsonSumCodec, which encodes sealed traits as illustrated above.

JsonSumCodec.typeField is provided as an alternative, which discriminates the various cases of a sealed trait by looking at a field, type, like

import argonaut._, Argonaut._, ArgonautShapeless._
import argonaut.derive._
implicit def typeFieldJsonSumCodecFor[S]: JsonSumCodecFor[S] =
  JsonSumCodecFor(JsonSumCodec.typeField)

sealed trait Base
case class First(i: Int) extends Base
case class Second(s: String) extends Base

val f: Base = First(2)

f.asJson.nospaces
// instead of the default """{"First":{"i":2}}"""
f.asJson.nospaces == """{"type":"First","i":2}"""

Proper handling of custom codecs

Of course, if some of your types already have codecs (defined in their companion object, or manually imported), these will be given the priority over the ones derived by argonaut-shapeless, like

import argonaut._, Argonaut._, ArgonautShapeless._

case class Custom(s: String)

object Custom {
  implicit def encode: EncodeJson[Custom] =
    EncodeJson.of[String].contramap[Custom](_.s)
  implicit def decode: DecodeJson[Custom] =
    DecodeJson.of[String].map(Custom(_))
}
Custom("a").asJson.nospaces == """"a""""
""""b"""".decodeOption[Custom] == Some(Custom("b"))

JsonCodec for local ArgonautShapeless._ import

If you want the codec derivation to happen at a controlled spot in your code, you can use the JsonCodec annotation. A situation would be that you have custom codecs and want to make sure they're considered or you don't want to overgenerate Codecs for ADTs instances.

import argonaut._, Argonaut._

object instances {
  import ArgonautShapeless._
  import argonaut.derive.JsonCodec

  @JsonCodec sealed trait ADT
  case class First(i: Int) extends ADT
  case class Second(s: String) extends ADT
  object ADT // this one's required
}

import instances._

// fails
// val encodeFirst = EncodeJson.of[First]
// works
val encode = EncodeJson.of[Base]

You'll have to add the additional object after all the instances of a sealed trait, see #5.

To use the @JsonCodec annotation, add MacroParadise to your build.

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

refined module

argonaut-shapeless also has a module to encode/decode types from refined, allowing for some kind of validation at the type level.

Add it to your dependencies with

libraryDependencies += "com.github.alexarchambault" %% "argonaut-refined_6.3" % "1.3.0"

Use like

import argonaut._, Argonaut._, ArgonautShapeless._, ArgonautRefined._
import eu.timepit.refined._
import eu.timepit.refined.api.Refined

case class CC(
  i: Int Refined numeric.Greater[W.`5`.T],
  s: String Refined string.StartsWith[W.`"A"`.T]
)

CC(
  refineMV(6),
  refineMV("Abc")
).asJson.nospaces == """{"i":6,"s":"Abc"}""" // fields are encoded as their underlying type

"""{"i": 7, "s": "Abcd"}""".decodeOption[CC] == Some(CC(refineMV(7), refineMV("Abcd")))
"""{"i": 4, "s": "Abcd"}""".decodeOption[CC] == None // fails as the provided `i` doesn't meet the predicate ``GreaterThan[W.`5`.T]``

See also

  • spray-json-shapeless features automatic codec derivation with shapeless for spray-json
  • circe features its own automatic codec derivation with shapeless, via its generic module

Contributors

Initially based on an early (non Lazy-based) automatic codec derivation in argonaut by Maxwell Swadling, Travis Brown, and Mark Hibberd.

License

Released under the BSD license. See LICENSE file for more details.