Feature request: add "null" default for optional fields

[erikvanoosten]

Most of the fields in my schema are of type Option. I would rather not specify the default (None) in the case class as I want users to explicitly give a value for each field. It is too easy to forget giving a value. Also, it is a lot of typing :)

It would be great if the generated schema could add a default "null" for fields that are a union of "null" and something else (provided they don't have some other default).

[erikvanoosten]

As a workaround you can use this code that updates the generated schema.

Use it like this:

import MoreAvroSupport._

case class Person(name: Option[String])

val PersonSchema: Schema = AvroSchema[Person].withOptionDefaults

import org.apache.avro.Schema

import scala.jdk.CollectionConverters._

object MoreAvroSupport {

  def withOptionDefaults(schema: Schema): Schema = {
    schema.getType match {
      case Schema.Type.RECORD =>
        val updatedFields = schema.getFields.asScala.map { field =>
          val newDefault =
            if (
              field.schema().isUnion &&
              field.schema().getTypes.asScala.headOption.exists(_.getType == Schema.Type.NULL) &&
              !field.hasDefaultValue
            ) {
              Schema.Field.NULL_DEFAULT_VALUE
            } else {
              field.defaultVal()
            }
          new Schema.Field(
            field.name(),
            withOptionDefaults(field.schema()),
            field.doc(),
            newDefault,
            field.order()
          )
        }.asJava
        val updatedSchema = Schema.createRecord(
          schema.getName,
          schema.getDoc,
          schema.getNamespace,
          schema.isError,
          updatedFields
        )
        schema.getAliases.asScala.foreach(updatedSchema.addAlias)
        schema.getObjectProps.asScala.foreach { case (k, v) => updatedSchema.addProp(k, v) }
        updatedSchema
      case Schema.Type.MAP =>
        val updatedValueType = Schema.createArray(withOptionDefaults(schema.getValueType))
        schema.getObjectProps.asScala.foreach { case (k, v) => updatedValueType.addProp(k, v) }
        Schema.createMap(updatedValueType)
      case Schema.Type.UNION =>
        val newTypes = schema.getTypes.asScala.map(withOptionDefaults).asJava
        Schema.createUnion(newTypes)
      case Schema.Type.ARRAY =>
        val updatedElementType = Schema.createArray(withOptionDefaults(schema.getElementType))
        schema.getObjectProps.asScala.foreach { case (k, v) => updatedElementType.addProp(k, v) }
        updatedElementType
      case _ => schema
    }
  }

  implicit class SchemaWithOptionDefaults(val schema: Schema) {

    /** @return
     *   `schema` updated to have "null" as default for any union type that contains the "null"
     *   type and doesn't have a default already.
     *
     * To get a default value in the schema Avro4s requires you to write case classes with an
     * explicit default value, like this:
     * {{{
     * case class Person(name: Option[String] = None)
     * }}}
     * This method adds those "null" defaults to the generated schema so you can write case classes
     * like this:
     * {{{
     * case class Person(name: Option[String])
     * }}}
     */
    def withOptionDefaults: Schema = MoreAvroSupport.this.withOptionDefaults(schema)
  }

}

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.