如何在 Scala 中将特征序列化和反序列化到 Json 或从 Json 反序列化？

Question

第一次尝试：

到目前为止我已经尝试过喷雾-json。我有：


trait Base

case class A ( id: String) extends Base

case class B (id: String) extends Base

现在，为了序列化和反序列化我的 Base 类型，我有代码：

implicit object BaseFormat extends RootJsonFormat[Base]{
    def write(obj: Base): JsValue = {
      obj match {
        case a: A => a.toJson
        case b: B => b.toJson
        case unknown @ _ => serializationError(s"Marshalling issue with ${unknown}")
      }
    }

    def read(json: JsValue): Base = {
      //how to know whether json is encoding an A or a B?
    }
  }

问题是，为了实现反序列化的 read 方法，我想不出一种方法来知道 JsValue 编码的是 A 还是 B .

第二次尝试：

为了在 spray-json 中解决这个问题，我最终只是将 A 中的字段 id 重命名为 aID，并将 B 中的字段重命名为 bID.

第三次尝试：

由于 spray-json 不如 zio-json 或 circe 等替代库复杂，它们无需额外代码即可自行处理此问题，因此我开始使用 zio-json

现在我得到了错误

magnolia: could not infer DeriveJsonEncoder.Typeclass for type

对于所有情况类采用类型参数。此外，它在链式特征继承方面存在问题。似乎 circe 也使用木兰。所以这很可能也会被 circe 复制。

如有任何帮助，我们将不胜感激。

Answer 1

您应该使用 json encoding/decoding 库解决这个问题。这是一个使用 circe 的例子，它是半自动模式。

既然你在评论中提到你在你的案例中遇到泛型类型问题 classes，我也把它包括在内。基本上，要为包含 T 的 class Foo[T] 导出编码器或解码器，您必须证明存在编码和解码 T 的方法。这是通过请求一个隐式的 Encoder[T] 和 Decoder[T] 来完成的，您可以在其中导出 Encoder[Foo[T]] 和 Decoder[Foo[T]]。当然，您可以将此推理概括为适用于多个泛型类型，您只需要隐含地使用一对 encoder/decoder，在其中为相应的情况 class 派生 encoder/decoder。

import io.circe._, io.circe.generic.semiauto._, io.circe.syntax._

case class Foo[T](a: Int, b: String, t: T)
object Foo {
  implicit def decoder[T: Encoder: Decoder]: Decoder[Foo[T]] = deriveDecoder
  implicit def encoder[T: Encoder: Decoder]: Encoder[Foo[T]] = deriveEncoder
}

case class Bar(a: Int)
object Bar {
  implicit val encoder: Encoder[Bar] = deriveEncoder
  implicit val decoder: Decoder[Bar] = deriveDecoder
}

case class Baz(a: Int)

println(Foo(42, "hello", 23.4).asJson) // Works because circe knows Encoder[Float] 
println(Foo(42, "hello", Bar(42)).asJson) // Works because we defined Encoder[Bar] 

//println(Foo(42, "hello", Baz(42)).asJson) // Doesn't compile: circe doesn't know Encoder[Baz] in semi-auto mode

Try it live on Scastie

请注意，需要为您使用的每种类型 T 生成 Foo[T] 的不同 encoder/decoder，这就是 [=21= 的编码器和解码器派生的原因] 必须是方法，而不是像 Bar.

这样的值

还有一个全自动模式，但它往往会产生编译时错误，对于初学者来说更难调试，所以我会从半自动开始。另一个问题是，自动模式在大型项目上编译可能需要更长的时间。如果你喜欢冒险，不犯错就更美了！

import io.circe._, io.circe.generic.auto._, io.circe.syntax._

case class Foo[T](a: Int, b: String, t: T)
case class Bar(a: Int)
case class Baz(a: Int)

println(Foo(42, "hello", 23.4).asJson) // circe knows Encoder[Float] and automatically derives Encoder[Foo[Float]]
println(Foo(42, "hello", Bar(42)).asJson) // circe wants Encoder[Foo[Bar]], so it ma(cro)gically derives Encoder[Bar] and then Encoder[Foo[Bar]]
println(Foo(42, "hello", Baz(42)).asJson) // Does compile this time, circe automatically find the encoder/decoder for Baz the same way it does for Bar

Try it live on Scastie

Answer 2

一个有效的 example 将 traits 与泛型结合使用是这样的：

import io.circe.Decoder.Result
import io.circe._
import io.circe.generic.semiauto._
import io.circe.syntax._


trait MyBase[T <: MyBase[T]] {
  def myid: String
  def issue: MyBaseIssue
  def summary: MyBaseSummary[T]
}

trait MyBaseIssue

object MyBaseIssue {
  implicit val decodeIssue: Decoder[MyBaseIssue] =
    Decoder[SimpleBaseIssue].map[MyBaseIssue](identity).or(
      Decoder[SophisticatedBaseIssue].map[MyBaseIssue](identity)
    )

  implicit val encodeIssue: Encoder[MyBaseIssue] = Encoder.instance {
    case simple @ SimpleBaseIssue(_) => simple.asJson
    case sophisticated @ SophisticatedBaseIssue(_) => sophisticated.asJson

  }
}

trait MyBaseSummary[T <: MyBase[T]]

object MyBaseSummary {
  implicit def decodeSummary[T <: MyBase[T]: Encoder: Decoder]
  : Decoder[MyBaseSummary[T]] =
    Decoder[SimpleBaseSummary].map[MyBaseSummary[SimpleBase]](identity).asInstanceOf[Decoder[MyBaseSummary[T]]].or(
      Decoder[SophisticatedBaseSummary].map[MyBaseSummary[SophisticatedBase]](identity).asInstanceOf[Decoder[MyBaseSummary[T]]]
    )


  implicit def encodeSummary[T <: MyBase[T]: Encoder: Decoder]
  : Encoder[MyBaseSummary[T]] = Encoder.instance {
    case simple @ SimpleBaseSummary(_) => simple.asJson
    case sophisticated @ SophisticatedBaseSummary(_) => sophisticated.asJson

  }
}

case class SimpleBase(
                       myid: String,
                       override val issue: SimpleBaseIssue,
                       override val summary: SimpleBaseSummary
                     ) extends MyBase[SimpleBase]

case class SophisticatedBase(
                              myid: String,
                              extraField: String,
                              override val issue: SophisticatedBaseIssue,
                              override val summary: SophisticatedBaseSummary
                            ) extends MyBase[SophisticatedBase]

object SimpleBase {
  implicit val encoder: Encoder[SimpleBase] = deriveEncoder
  implicit val decoder: Decoder[SimpleBase] = deriveDecoder
}

object SophisticatedBase {
  implicit val encoder: Encoder[SophisticatedBase] = deriveEncoder
  implicit val decoder: Decoder[SophisticatedBase] = deriveDecoder
}

case class SimpleBaseIssue(simpleIssueType: String) extends MyBaseIssue
case class SophisticatedBaseIssue(sophisticatedIssueType: String) extends MyBaseIssue

object SimpleBaseIssue {
  implicit val encoder: Encoder[SimpleBaseIssue] = deriveEncoder
  implicit val decoder: Decoder[SimpleBaseIssue] = deriveDecoder
}

object SophisticatedBaseIssue {
  implicit val encoder: Encoder[SophisticatedBaseIssue] = deriveEncoder
  implicit val decoder: Decoder[SophisticatedBaseIssue] = deriveDecoder
}

case class SimpleBaseSummary(simpleSummary: String) extends MyBaseSummary[SimpleBase]

case class SophisticatedBaseSummary(sophisticatedSummary: String) extends MyBaseSummary[SophisticatedBase]

object SimpleBaseSummary {
  implicit val encoder: Encoder[SimpleBaseSummary] = deriveEncoder
  implicit val decoder: Decoder[SimpleBaseSummary] = deriveDecoder
}

object SophisticatedBaseSummary {
  implicit val encoder: Encoder[SophisticatedBaseSummary] = deriveEncoder
  implicit val decoder: Decoder[SophisticatedBaseSummary] = deriveDecoder
}

case class MyBaseList[T <: MyBase[T]](
                                       myid: String,
                                       var membersMap: Map[String, T] = Map.empty,
                                       override val issue: MyBaseListIssues[T],
                                       override val summary: MyBaseListSummary[T]
                                     ) extends MyBase[MyBaseList[T]]

object MyBaseList {

  implicit def membersMapDecoder[T <: MyBase[T]: Encoder: Decoder]
  : Decoder[Map[String, T]] = Decoder.decodeMap
  implicit def membersMapEncoder[T <: MyBase[T]: Encoder: Decoder]
  : Encoder[Map[String, T]] = Encoder.encodeMap


  implicit def encoder[T <: MyBase[T]: Encoder: Decoder]
  : Encoder[MyBaseList[T]] = deriveEncoder
  implicit def decoder[T <: MyBase[T]: Encoder: Decoder]
  : Decoder[MyBaseList[T]] = deriveDecoder
}

case class MyBaseListIssues[T <: MyBase[T]](
                                             issue: String,
                                             var set: Set[MyBaseIssue] = Set[MyBaseIssue]()
                                           ) extends MyBaseIssue

object MyBaseListIssues {
  implicit def membersMapDecoder: Decoder[Set[MyBaseIssue]] =
    Decoder.decodeSet[MyBaseIssue]
  implicit def membersMapEncoder: Encoder[Set[MyBaseIssue]] = Encoder.encodeSet

  implicit def encoder[T <: MyBase[T]: Encoder: Decoder]
  : Encoder[MyBaseListIssues[T]] = deriveEncoder
  implicit def decoder[T <: MyBase[T]: Encoder: Decoder]
  : Decoder[MyBaseListIssues[T]] = deriveDecoder
}

case class MyBaseListSummary[T <: MyBase[T]](
                                              summaryList: Map[String, MyBaseSummary[T]]
                                            ) extends MyBaseSummary[MyBaseList[T]]

object MyBaseListSummary {

  implicit def membersMapDecoder[T <: MyBase[T]: Encoder: Decoder]
  : Decoder[Map[String, MyBaseSummary[T]]] = Decoder.decodeMap
  implicit def membersMapEncoder[T <: MyBase[T]: Encoder: Decoder]
  : Encoder[Map[String, MyBaseSummary[T]]] = Encoder.encodeMap


  implicit def encoder[T <: MyBase[T]: Encoder: Decoder]
  : Encoder[MyBaseListSummary[T]] = deriveEncoder
  implicit def decoder[T <: MyBase[T]: Encoder: Decoder]
  : Decoder[MyBaseListSummary[T]] = deriveDecoder
}



object MainObject extends App {
  val simpleList = MyBaseList[SimpleBase]("simpleId",
    Map("one" -> SimpleBase("baseid", SimpleBaseIssue("the mistake"), SimpleBaseSummary("very concise"))),
    MyBaseListIssues[SimpleBase]("listIssue", Set(SimpleBaseIssue("a disaster"))),
    MyBaseListSummary[SimpleBase]( Map("simplebaseid" -> SimpleBaseSummary("super concise")))
  )
  val simpleJson = simpleList.asJson
  println(simpleJson)
  val convertedSimpleList = simpleJson.as[MyBaseList[SimpleBase]]
  println(convertedSimpleList)


  val sphisticatedList = MyBaseList[SophisticatedBase]("sophisticatedId",
    Map("one" -> SophisticatedBase("baseid", "further detail", SophisticatedBaseIssue("the mistake"), SophisticatedBaseSummary("very concise"))),
    MyBaseListIssues[SophisticatedBase]("listIssue", Set(SophisticatedBaseIssue("a disaster"))),
    MyBaseListSummary[SophisticatedBase]( Map("sophisticatedbaseid" -> SophisticatedBaseSummary("super concise")))
  )
  val sophisticatedJson = sphisticatedList.asJson

  println(sophisticatedJson)

  val convertedSophisticatedListDecoder: Result[MyBaseList[SophisticatedBase]] = sophisticatedJson.as[MyBaseList[SophisticatedBase]]

  println(convertedSophisticatedListDecoder)

  convertedSophisticatedListDecoder match {
    case Left(failure) => println(failure)
    case Right(list) => println(list)
  }
}

仍然必须确保每个继承特定特征的案例 class 与继承相同特征的其他案例具有不同数量的字段，或者至少与其他案例不同 classes 在这些字段的命名中。

否则，解码器将无法正确完成其工作并将 json 解码为在解码器函数中匹配它的第一种情况 class，而不是正确的预期情况。

如何在 Scala 中将特征序列化和反序列化到 Json 或从 Json 反序列化？

How to serialize and deserialize traits, to and from Json, in Scala?

json

scala

spray-json

circe