fix: ignore __type field when deserializing unions in json protocols

[0marperez]

Issue #

closes aws-sdk-kotlin#1044

Description of changes

-Added a flag for JSON protocols
-Added a __type to union object descriptors
-Ignoring __type unless it's member explicitly in union shape

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[ianbotsf]

I think we should move more of this custom logic into the the JSON-specific classes. Burying this in abstract or base implementations defeats the purpose of polymorphism and makes the code harder to follow.

There's a bunch of ways this could be done. One way, I think, is the descriptor logic could be moved into JsonSerdeDescriptorGenerator, which could manually add a __type item to the memberShapes list. The logic for rendering the field deserializer could be moved into a JSON-specific union deserializer generator which inherits DeserializeUnionGenerator and overrides renderShapeDeserializer.

[sonarqubecloud]

SonarCloud Quality Gate failed.   

0 Bugs
0 Vulnerabilities
0 Security Hotspots
9 Code Smells

No Coverage information
71.8% Duplication

Catch issues before they fail your Quality Gate with our IDE extension SonarLint

[aajtodd]

These changes are to account for AWS specific protocol behaviors and as such mostly should be implemented in aws-sdk-kotlin codegen.

I also think we can simplify this and not need to special case __type as a pseudo field. Instead we can just define a new deserialization trait to ignore it all together and support that in the runtime deserializer.

e.g.

  val X_DESCRIPTOR = SdkFieldDescriptor(SerialKind.Integer, JsonSerialName("x"))
  val Y_DESCRIPTOR = SdkFieldDescriptor(SerialKind.Integer, JsonSerialName("y"))
  val OBJ_DESCRIPTOR = SdkObjectDescriptor.build {
      field(X_DESCRIPTOR)
      field(Y_DESCRIPTOR)
      trait(IgnoreKeys("__type"))   // new object trait to signal that `__type` should just be ignored and not even enumerated in the `findNextFieldIndex()` logic.
  }


deserializer.deserializeStruct(OBJ_DESCRIPTOR) {
    loop@ while (true) {
        when (findNextFieldIndex()) {
            X_DESCRIPTOR.index -> x = value = MyUnion.X(deserializeInt())
            Y_DESCRIPTOR.index -> y = MyUnion.Y(deserializeInt())
            null -> break@loop
            else -> value MyUnion.SdkUnknown.also { skipValue() }
        }
    }
}

This should require way less changes in codegen since the AWS protocols just need to override JsonSerdeDescriptorGenerator.getObjectDescriptorTraits() and be relatively easy to add to to the JsonDeserializer in the runtime.

[aajtodd]

fix: This logic is specific to AWS protocols and doesn't belong in smithy-kotlin

[aajtodd]

These tests aren't really providing any additional value (especially since this behavior is covered by protocol tests anyway).