zio / caliban-deriving   0.0.1

Apache License 2.0 Website GitHub

Full-featured, robust deriving for Caliban.

Scala versions: 3.x 2.13 2.12

Caliban Deriving

Full-featured, robust deriving for Caliban.

Development CI Badge Sonatype Releases Sonatype Snapshots javadoc Caliban Deriving

The Caliban Deriving library provides an advanced derive macro for the Caliban library that seamlessly integrates with the built-in schema derivation system and allows the library users to not only derive the schema for the user defined case classes and enums but also for calculated fields and functions with parameters.

Getting started

Start by adding caliban-deriving as a dependency to your project:

libraryDependencies += "dev.zio" %% "caliban-deriving" % "0.0.1"

Once the library is added, Caliban's default auto-derived schemas can be replaced one by one for each type by explicitly deriving its schema using the Caliban Deriving method:

case class Example(value: Int)
object Example {
    implicit lazy val exampleSchema: Schema[Any, Example] =
      deriveSchemaInstance[Any, Example]
}

Please read Caliban's official documentation about schemas to learn more about what a schema is and what features the core library provides.

It is important that it is possible to use a mix of Caliban's built-in schemas, auto-derived schemas and the ones provided by deriveSchemaInstance. Both derive methods are looking for implicit instances of Schema.

Features

Caliban Deriving's deriveSchemaInstance function can be applied on the following Scala data types:

  • case classes
  • sealed traits or Scala 3 enums
  • any trait

Common features

The following rules apply equally for case classes, sealed traits and traits.

  • public val fields and parameterless def methods are generated as GraphQL fields. Their Schema is found by an implicit search for the given type. This way it supports exactly the same cases as Caliban's built-in derive method, including fields of type ZIO, ZQuery and Future.
  • public methods with one or more parameters are generated as GraphQL functions. The Schema of the function arguments is found by an implicit search for their type.
  • protected and private fields/methods are excluded
  • public members annotated by @GQLExclude are also excluded
  • the following core Caliban annotations are supported: @GQLName, @GQLDescription, @GQLDeprecated and @GQLDirective

Case classes

From case classes a GraphQL type is generated. If it is used as an input (as a parameter for a function in another type), then an GraphQL input is generated which only contains the constructor parameters of the case class, not the other members.

Sealed traits

For sum types (sealed trait with a set of case class / case objects implementations) if the base trait contains any methods, it will become a GraphQL interface and each constuctor a type.

If the base trait has no members, but the constructors have parameters, it becomes a GraphQL union. Members of constructors are still handled the same way as mentioned above.

If all the constructors are case objects then it becomes a GraphQL enum.

Traits

If the derivation is invoked on a simple trait (not sealed) then it works as if it were a concrete parameterless case class. This means a GraphQL type is generated and a schema derived for an arbitrary trait can be used to serve any implementation of it.

Using with effects

When there are members of type ZIO or ZQuery, they can have an environment parameter (-R) that is not Any. Effects and queries without environment just work.

When working with effects requiring an environment the following rules apply:

  • You have to have the GenericSchema[R] trait in scope as described in the official documentation. This should be done by inheriting this trait in an object and putting all the additional derived implicits there.
  • The implicit resolution in this case will only be correct if you put all the implicit instances derived with Caliban Deriving's deriveSchemaInstance directly into the object extending GenericSchema.
  • Although the macro can calculate the union of all the required environments used in a type, it is not automatically using it for the result Schema. The reason is that you may want to derive schema for multiple data types, each using only a subset of the total environment passed to the GenericSchema. In this case the full environment cannot be calculated inside the derivation. For this reason, you must provide the full required environment, the same that is passed to the GenericSchema in scope, to the deriveSchemaInstance method in its first type parameter.

Mutations and subscriptions

Not supported yet.

Documentation

Learn more on the Caliban Deriving homepage!

Contributing

For the general guidelines, see ZIO contributor's guide.

Code of Conduct

See the Code of Conduct

Support

Come chat with us on Badge-Discord.

License

License