[oryan-block]

Description

Fixes #938

A generic wrapper registered with a transformer that returns a future of its content, like Reactor's Mono with Mono::toFuture, lost a future when its type argument was itself a future. For Mono<CompletableFuture<Item>>, { item { name } } failed with

Expected source object to be an instance of 'Item' but instead got 'java.util.concurrent.CompletableFuture'

and a Mono<CompletionStage<String>> field returned "java.util.concurrent.CompletableFuture@4d3c6593[Completed normally]" with no error. Same inside lists and other wrappers.

The unwrapper hands a wrapper's transformer result straight to the declared future below it. That's right for a Try<CompletableFuture<Item>> whose transformer returns the declared future (graphql-java gets the resolver's own future), but Mono::toFuture returns a future of it. Both build the same steps, and only the completed value tells them apart. Checking that would add a thenCompose on every wrapper around a future and misread union members that implement Future, so the wrapper has to say it.

Now:

• GenericWrapper.withCompletionStageTransformer(Mono::class, 0, Mono<*>::toFuture) (or (Mono.class, 0, Mono::toFuture) from Java) registers a wrapper whose transformer returns a CompletionStage of its content. The unwrapper waits on that future first, then on the declared one, at the top level and nested
• wrappers registered with withTransformer behave as before, so a Mono registered that way still has the bug. Nothing is guessed at runtime
• GenericWrapper itself is unchanged (constructor, copy, componentN). The factory marks the transformer instead, and copy keeps it since it keeps the transformer

I first had the flag as a new GenericWrapper property. Keeping the old JVM signatures then needs a hand-written 4-parameter constructor and copy that read as dead code, and a caller could set the flag on a transformer that returns the content. With the factory, the transformer's type has to be a CompletionStage.

No DataFetchingEnvironment or schemaWrapper overloads, since nobody asked for them. Easy to add later.

Release note: GenericWrapper.withCompletionStageTransformer registers a wrapper whose transformer returns a future of its content, such as Reactor's Mono::toFuture. A future declared inside it, like Mono<CompletableFuture<Item>>, is then waited on too. Before, graphql-java got the inner future as the value. Wrappers registered with withTransformer keep their behaviour.

Checklist

• Pull requests follows the contribution guide
• New or modified functionality is covered by tests

mvn clean verify passes (440 tests). New tests in NestedGenericWrapperTest cover the future at the top level, a scalar, a list, and a union member that implements Future. The first three fail without the unwrapper change.

Also checked through the installed jar in a separate consumer project with reactor-core 3.7.19. With withCompletionStageTransformer, Mono<CompletableFuture<Item>>, List<Mono<CompletableFuture<Item>>>, Mono<CompletionStage<String>> and Try<Mono<CompletableFuture<Item>>> all resolve. With withTransformer they still fail like on master. A Java caller compiles with -Xlint:all -Werror (apart from the raw Mono.class), and javap against 14.5.0 shows no removed or changed GenericWrapper signature.

The bounded check of every return type up to 3 containers deep (about 172k executions), with its future-returning wrapper registered through the new factory, no longer finds the 192 failing return types it finds on master, and finds nothing new.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud