Ignore namespace

[geny200]

Based on #3 issue - it was closed, but was added only ability to delete Namespace, it's not "true" ignore (because if the Namespace is specified in the model in the code, it'll still be required during decoding). My case: I have a wrapper that adds nesting and provides Namespace, and I need to be able to decode and encode. In the case of encoding, I want the resulting xml to have Namespace, in the case of deserialization, I want it to be ignored.

Code example:

//> using dep dev.valentiay::phobos-core:0.26.0
//> using scala "2.13.16"

import phobos.Namespace
import phobos.configured.ElementCodecConfig
import phobos.decoding.{ElementDecoder, XmlDecoder}
import phobos.derivation.semiauto.{deriveElementDecoder, deriveElementEncoder, deriveXmlDecoderConfigured}
import phobos.encoding.{ElementEncoder, XmlEncoder}
import phobos.syntax.xmlns

object MyNamespace {
  type ns = MyNamespace.type
  implicit val ns: Namespace[ns] = Namespace.mkInstance[ns]("http://test.org/some", Some("my"))
}

final case class Wrapper[T](@xmlns(MyNamespace) foo: T)

object Wrapper {
  implicit def encoder[T: ElementEncoder]: ElementEncoder[Wrapper[T]] = deriveElementEncoder
  implicit def decoder[T: ElementDecoder]: ElementDecoder[Wrapper[T]] = deriveElementDecoder

  implicit def xmlEncoder[T: ElementEncoder]: XmlEncoder[Wrapper[T]] =
    XmlEncoder.fromElementEncoder[Wrapper[T]]("MyRoot")
  implicit def xmlDecoder[T: ElementDecoder]: XmlDecoder[Wrapper[T]] =
    deriveXmlDecoderConfigured[Wrapper[T]](
      "MyRoot",
      ElementCodecConfig.default.withRemoveNamespaces, // remove namespaces
    )
}

object Main extends App {

  // case 1
  println(XmlEncoder[Wrapper[String]].encodeUnsafe(Wrapper("bar")))
  // result: <?xml version='1.0' encoding='UTF-8'?><MyRoot><my:foo xmlns:my="http://test.org/some">bar</my:foo></MyRoot>

  // case 2
  println(XmlDecoder[Wrapper[String]].decode(
    """<?xml version='1.0' encoding='UTF-8'?><MyRoot><my:foo xmlns:my="http://leviy-soft.org/more">bar</my:foo></MyRoot>"""
  ))
  // result: Left(phobos.decoding.DecodingError: Error while decoding XML: Invalid namespace. Expected 'http://test.org/some',  but found ''
  //	In element 'foo'
  //		in element 'MyRoot'
  //     )

}

In second case I expects as a result Wrapper("bar")

[valentiay]

Hi!

It's not possible to ignore namespaces in phobos - element namespace validation is currently built in ElementDecoder instances, so it is not possible to configure this check.

The withRemoveNamespaces settings allows to bypass this check if namespace validation is not desired. It works as if the document was first transformed from

<?xml version='1.0' encoding='UTF-8'?><MyRoot><my:foo xmlns:my="http://leviy-soft.org/more">bar</my:foo></MyRoot>

to

<?xml version='1.0' encoding='UTF-8'?><MyRoot><foo>bar</foo></MyRoot>

and then it passes this document to XmlDecoders and ElementDecoders.

So you have two options:

1. Use withRemoveNamespaces and remove @xmlns annotations from code. In this case you may need different case classes for encoding and decoding.
2. Not use withRemoveNamespaces, but specify correct namespaces in @xmlns or with elementsDefaultNamespace setting.

I'll try to add a convenient way to ignore namespaces in phobos 1.0, but it won't come out any time soon

[geny200]

you may need different case classes for encoding and decoding

Actually it's the reason, why I need it. I don't want to boilerplate code for models

I'll try to add a convenient way to ignore namespaces in phobos 1.0, but it won't come out any time soon

Thx)