softwaremill / softwaremill/tapir

Support for error ADT and circe Codec in output

Open
#1,043 6 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Scala
Stars
1.5k
Forks
468
Avg merge
5h 37m
Merged PRs (30d)
34

Description

Tapir version: 0.17.12

Scala version: 2.13.5

Prerequisites

So let's say I have the following error ADT:

sealed trait ErrorInfo

case class NotFound(what: String) extends ErrorInfo
case class Unauthorized(realm: String) extends ErrorInfo
case class Unknown(code: Int, msg: String) extends ErrorInfo
case object NoContent extends ErrorInfo

object ErrorInfo {
   implicit val circeCodec: circe.Codec[ErrorInfo] = ???
}

And defined circe codec produces and expects the discriminator object:

NotFound(what = "something").asJson.spaces2 == 
"""
{   
  "not_found" : {
      "what" : "something"
   }
}
"""

Problem

I want to use ErrorInfo in the error output of my endpoint, so I write this code:

val baseEndpoint = endpoint.errorOut(
  oneOf[ErrorInfo](
    statusMapping(StatusCode.NotFound, jsonBody[NotFound].description("not found")),
    statusMapping(StatusCode.Unauthorized, jsonBody[Unauthorized].description("unauthorized")),
    statusMapping(StatusCode.NoContent, emptyOutput.map(_ => NoContent)(_ => ())),
    statusDefaultMapping(jsonBody[Unknown].description("unknown"))
  )
)

And it does not compile, because the compiler can't find circe codec for each case of the ADT:

[error]  could not find implicit value for evidence parameter of type io.circe.Encoder[NotFound]
[error]           statusMapping(StatusCode.NotFound, jsonBody[NotFound].description("not found")),

which is understendable actually because of jsonBody signature: def jsonBody[T: Encoder : Decoder: Schema] — T here and in circe's typeclasses is invariant.

I could use import io.circe.generic.auto._ but then the output differs from what I expect: it's just plain JSON without discriminator: { "what" : "something" }.

Workaround

For now, I am using this custom output:

 def jsonBodyADT[A: Encoder: Decoder, B <: A: Schema: ClassTag]: EndpointIO.Body[String, B] = {
    implicit val bEncoder: Encoder[B] = Encoder.instance[B](Encoder[A].apply(_))
    implicit val bDecoder: Decoder[B] = Decoder.instance[B](
      json =>
        Decoder[A].apply(json) match {
          case Left(value) => Left(value)
          case Right(value) if value.getClass == implicitly[ClassTag[B]].runtimeClass =>
            Right(value.asInstanceOf[B]) //scalafix:ok
          case _ => Left(DecodingFailure.fromThrowable(new Throwable("wrong subtype"), Nil))
        }
    )
    
 //and then in oneOf(...)
    statusMapping(StatusCode.NotFound, jsonBodyADT[ErrorInfo, NotFound].description("not found"))

and this works but feels very wrong.

Question

So what can be done to fix this behaviour? Is there a built-in way to work with that in Tapir? If there is none, can I contribute it?

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by examining the jsonBody typeclass requirements and the oneOf/statusMapping output definitions described in the issue. Compare them with the jsonBodyADT workaround and the discriminator-shaped circe codec. Done should mean an ErrorInfo endpoint output can use subtype mappings without separate subtype codecs while preserving the expected discriminator JSON.

Written by the indexing model from the issue text.

Assessment

Tech stack
scala
Domain
api, backend-api-design
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.