graph_wrap VS zimagi

This doesn't directly answer the question, but I wrote a library which uses meta programming to build a GraphQL Graphene API from a DRF API https://github.com/PaulGilmartin/graph_wrap. Maybe it's the type of thing which would interest you.

Enter GraphQL: GraphQL is designed so that the client decides what info it receives from the server, not the other way around. Whilst many great packages exist to create a GraphQL API from scratch, migrating an mature production REST API to use one of these frameworks is not so simple. It may also be that our REST API has functionality which is not available on a GraphQL specific API. This is where GrapWrap comes in: by adding two lines of code to your project, GraphWrap exposes a GraphQL schema which has the same "shape" as your existing REST API. With this new endpoint, we can now stop overexposing the author fields and instead simply expose author as a URL:

I attempt to give a use case for user defined queries here: https://github.com/PaulGilmartin/graph_wrap#which-problems-does-graphwrap-address

See Which problems does GraphWrap address for more.

Happy to share an example from an open source project I am working on in Django (https://github.com/zimagi/zimagi). You can find the actual API generation logic in the {app/systems/api} folder if you are interested. Basically what I do is declare all models as YAML declarations then generate the model classes and API endpoints, filters, and serializer fields based off of that YAML so I can bypass coding the APIs.