paulcwatts / drf-json-schema Goto Github PK

http://json-schema.org/
Your just using the same name...?

Since Django 3.8, a Django application needs to have a full Python path and unique across the project. [1]

Basic repro (write up a unit test):

1. Create a list serializer that attaches some links to the data
2. Wrap the serializer in a viewset with pagination
3. Expect that the links appear in the rendered output.

What actually happens is they are ignored, because the JSONAPILimitOffsetPagination class ignores them when it wraps the data in a paginated output.

Perhaps consider an easier way to attach links, like allowing UrlLink() objects in a subclass of the renderer and generic 'render_links' or 'render_meta' in the renderer, similar to the way we allow for ResourceObjects. Alternatively, try to find a way to get something like this from the schema? That would centralize all of the links and meta handling in the schema, which would be useful.

Currently a serializer needs to specify a schema in order for the renderer to work. Wouldn't it be nice if it generated a schema (like what DRF does with coreapi) if none was associated? Let me know if this aligns with the project goals and if there are any special considerations to be made.

If you set a serializer field as write_only, DRF doesn't include it in the serializer data. However, the Schema object expects that all fields that are specified in the attributes and relationships lists appear in the serializer data. This causes a KeyError to be thrown in Schema.render_attributes.

Support write_only attributes by ignoring the attribute or relationship if it's not in the data.