Add TypedDict for params and data arguments; add type hint for kwargs

[DXsmiley]

Here's the changes that I actually wanted to make before getting caught up with the non-determinism lol

I was getting a lot of errors from my typechecker (pyright) while trying to use parts of the library (it doesn't like function types annotated with a raw dict), so I've replaced these with TypedDict types to more accurately reflect the values that the functions actually allow. They're approximately the same as the equivalent model types.

I've also annotated **kwargs on these functions with Any since the typechecker was complaining about it being unannotated, even when I wasn't supplying a value for it. It might be possible to more accurately type it with Unpack, however I think that's going to take enough effort to warrant a separate PR.

This change should only impact static type checkers and shouldn't impact runtime behaviour.

[MarshalX]

general quesiton: will it give the ability to pass arbitrary keys to such dicts without problems (i mean without typing errors)? for example user may want to create records with custom fields that are not described in lexicon but could be saved in repo

[MarshalX]

general question 2: pydantic stores all definitions (including resolved types) of fields inside each model. Maybe is it possible to reuse it for dicts somehow? just and idea

[MarshalX]

general question 2: pydantic stores all definitions (including resolved types) of fields inside each model. Maybe is it possible to reuse it for dicts somehow? just and idea

answer from https://stackoverflow.com/a/74769085/8032027:

If you are wondering, whether there is a way to just dynamically infer the TypedDict rather than just duplicating the model fields manually, the answer is no.

Okay :(

[DXsmiley]

general quesiton: will it give the ability to pass arbitrary keys to such dicts without problems (i mean without typing errors)? for example user may want to create records with custom fields that are not described in lexicon but could be saved in repo

It won't allow arbitrary keys. I wasn't aware this was a feature (of bluesky) at all. Can you point to an example in the codebase where someone might want to do this?

[MarshalX]

general quesiton: will it give the ability to pass arbitrary keys to such dicts without problems (i mean without typing errors)? for example user may want to create records with custom fields that are not described in lexicon but could be saved in repo

It won't allow arbitrary keys. I wasn't aware this was a feature (of bluesky) at all. Can you point to an example in the codebase where someone might want to do this?

For example this post record contains custom field https://github.com/MarshalX/atproto/blob/main/tests/models/tests/test_custom_post_record.py

And it was possible to create such record using sdk. Via dict in the args

[MarshalX]

here is how to create such records:

    import ....


    client = Client()
    client.login(...)

    created_post = client.com.atproto.repo.create_record(
        models.ComAtprotoRepoCreateRecord.Data(
            repo=client.me.did,
            collection=ids.AppBskyFeedPost,
            record={
                'createdAt': client.get_current_time_iso(),
                'text': 'custom record',
                'lol': 'kek',
                'lol2': 'kek2',
            },
        )
    )

    rkey = AtUri.from_str(created_post.uri).rkey
    post = client.com.atproto.repo.get_record(
        {'collection': ids.AppBskyFeedPost, 'repo': client.me.did, 'rkey': rkey}
    )

    print(post)

probably it will work with new changes because of record: 'UnknownType' where UnknownType: te.TypeAlias = t.Union[UnknownRecordTypePydantic, 'dot_dict.DotDictType']

But will pyright raise errors on the provided code above because of DotDictType (not typed dict)?

[DXsmiley]

I've added UnknownInputType, which allows plain dicts in places where a user might supply one. Result types are still the same though.

[DXsmiley]

done!

[MarshalX]

thank you!!