io.bullet.borer.derivation

Derivation macros for array-based encodings.

Same as MapBasedCodecs, but with "compact", i.e. unwrapped, encodings for unary case classes.

Allows for custom configuration of certain derivation behaviors.

Customizations must be brought into implicit scope for activation, e.g.

 implicit val myConfig = DerivationConfig(
   encodeCaseClassMemberDefaultValues = true
 )

Base class of all macro-generated AdtDecoders.

Base class of all macro-generated AdtEncoders.

Derivation macros for array-based encodings.

Annotation allowing for customizing

1. the id identifying which concrete subtype of an ADT is encoded/to be decoded
2. the name of case class members (de)serialized with MapBasedCodecs

USE CASE 1: Type Ids

By default the derived encoders and decoders for ADTs serialize the short class name of an ADT instance in order to let the decoder now, which subtype to deserialize.

This notation allows for customizing the type id that is used to identify an ADT sub-type in the encoding and offers two choices: A String or an integer number (Int/Long) value. (The latter is not supported in JSON and will cause any encoding attempt to fail with an exception!)

Relying on strings makes the serialization more descriptive but is more verbose on the wire. Relying on numbers makes the serialization very compact but less descriptive.

If necessary styles could also be mixed within a single ADT, even though this is not recommended (for consistency/clarity only, there is no technical reason).

The following example shows such an (otherwise not recommended) mixed setup for illustrating the different style of using this annotation for type ids:

 sealed trait Animal
 object Animal {
   @key("D") // default would be "Dog", but we want sth shorter
   final case class Dog(name: String, age: Int)

   @key(2)
   final case class Cat(name: String, age: Int, weight: Option[Double])

   // if no @TypeId is given, the short type name is used, in this case "Mouse"
   final case class Mouse(name: String, size: Int)
 }

USE CASE 2: Member Names

By default MapBasedCodecs use the case class member name as map key, which is usually a good default. (Note: This will also work with backtick-ed member names as in case class(`type`: Int, `foo-bar`: String)).

However, sometimes it's convenient to be able to map the case class member to another name (and vice versa). This annotation allows for just that. As already described above (USE CASE 1) the annotation can take a String or an integer number (Int/Long) value as key. Mixing the two different styles in one class is allowed but not recommended. JSON only supports String keys.

This annotation is ignored by ArrayBasedCodecs.

Example:

 case class Cat(name: String, age: Int, @key("weight") kilos: Option[Double])