graphql.language.Field usage

[momentary-lapse]

Hello! I have a couple of questions about the usage of graphql.language.Field class.

1. Is there a convenient way to inject something into the tree structure of the Field? Let's say I have something like that:

field {
  subfield1
  subfield2 {
    subsubfield1
    subsubfield2
  }
}

as a Field, and I want to insert subsubfield3 in the selection set of subfield2. I can do it using transform with quite a lot of boilerplate code, wondering if it's the only way.

2. AstPrinter allows to print the Field (and any other Node) into a String. Is the opposite possible? I'd like to parse a String into a Field.

I realize that maybe it's not a very common usage of the Field, so let me clarify why I'm asking this: I have different generic GraphQL requests which selectionSet I need to change, and send them to an underlying service. I have access to DataFetchingEnvironment corresponding to these requests, so I retrieve its Field and work with it.

Thanks in advance 🙏 If the functionality is missing but you would consider it useful, I would be able to contribute

[bbakerman]

graphql.language.Field as its name suggests is the AST representation of graphql query text. And yes it can be parsed into a string with some tricks

The parser does NOT start from a syntax element of field but its can be tricked

    static Field parseField(String sdlField) {
        String spec = """ query Foo {
        $sdlField
        }
        """
        def document = parseQuery(spec)
        def op = document.getDefinitionsOfType(OperationDefinition.class)[0]
        return op.getSelectionSet().getSelectionsOfType(Field.class)[0] as Field
    }

parseField("f(arg : 1) { inner sub selection }") for example

The thing is Field and graphql.language.SelectionSet is not really a representation of the "executable notion of a field and its subselection"

They are the AST syntax elements. For example read https://www.graphql-java.com/blog/deep-dive-merged-fields

{
   foo
   foo
}

This is valid AST and will include two graphql.language.Field objects but there is only one runtime field

You need some other representation of a "field and its sub selection". With the https://github.com/atlassian-labs/nadel (our Atlassian federated query engine) we use graphql.normalized.ExecutableNormalizedField and graphql.normalized.ExecutableNormalizedOperation - this is a more complete representation of the "runtime" when you take out the syntactical elements

You can mutate this structure and then "print" a sub document from this to form "sub operations"

If you are just wanting to know "the simple runtime field subselection" then graphql.schema.DataFetchingEnvironment#getSelectionSet and graphql.schema.DataFetchingFieldSelectionSet is what you want.

See https://www.graphql-java.com/blog/deep-dive-data-fetcher-results

[momentary-lapse]

Thanks much for the explanation and this code snippet!
Yeah, I'm aware of the merged fields in AST. We're expecting our API to be used only internally, so for now I need "the simple runtime field subselection", and access it with something likeDataFetchingEnvironment.getMergedField().getSingleField(). We're bounded to this DataFetchingEnvironment class because of DGS framework, their fetchers are exposing it for particular fields and operations. But will take a look into ExecutableNormalizedOperation as well.