Fields declared in subclasses are not recognized

[sblack-usu]

from schema_models.creative_work import CreativeWork

def test_creative_work_name():
    c = CreativeWork(awards="awardtest")

The above test results in

E       TypeError: Thing.__init__() got an unexpected keyword argument 'awards'

I haven't used the fquery.pydantic decorator before exploring this library, so my knowledge on it is limited. Based on the error, I'd guess all classes that are not base classesand are broken because they do not have the fquery.pydantic decorator. The decorator seems to only be only recognizing the base class when I use the subclasses.

Should all of the schema-org subclasses also use the fquery.pydantic decorator? I ran a few tests and it seemed to work when I did this. Is there any reason why the decorator isn't used on subclasses currently?

I also think it'd be worth documenting this repository isn't strictly pydantic classes, rather it is using fquery to build dataclass/pydantic classes.

I think this is an interesting and useful project. I'm interested in using it and contributing back to the library.

[adsharma]

I removed the decorator from subclasses while debugging a different problem since documentation said this is ok:

from pydantic import BaseModel

# Base model
class Person(BaseModel):
    name: str
    age: int

# Subclassed model with additional fields
class Employee(Person):
    job_title: str
    employee_id: int

All that the @pydantic decorator is doing is injecting BaseModel as a parent class.

But for @DataClass, both the base and derived class need the decorator. I tested that adding a decorator works ok.

Thank you for finding this. Will send a PR.

Would love to see more contributors in the future.

[adsharma]

Please test the linked fix.

[adsharma]

@sblack-usu did the PR work for you?

[sblack-usu]

I did take a look at it. The increase in recursion limit and startup time is a bit concerning, at least for that application I had in mind. I do think it'd likely be ok if you're doing a lot of searching.

[adsharma]

I share your concern, which is why I haven't merged the PR. I profiled the time spent at startup - it's all inside pydantic.

If there are any ideas on how the work can be done lazily instead of at startup, I'm very interested in incorporating.