The first thing I'd like to get some help is how to create nested field in a SchemaMetadata. As you can see from my screenshot, I can parse the first level of the fields, but I'm not sure about nested fields. I looked at HiveSource and TrinoSource, but I did not see what I was expecting. I was expecting to find code composing

datahub.metadata.schema_classes.*

types together, but instead was returning some kind of JSON:

return {
            "type": "map",
            "values": vt,
            "native_data_type": s,
            "key_type": kt,
            "key_native_data_type": parts[0],
        }

In my implementation, I'm strictly using Python classes like

NumberTypeClass

,

ArrayTypeClass

and

RecordTypeClass

, so now I'm wondering if I started on the wrong foot... Are Hive and Trino good examples to copy? They are both based on SqlAlchemy, so maybe not?