[0.16.0 regression] TypeError: Field names must be valid identifiers

[andersk]

As of commit c2e166e (#422), trying to validate {"0": "a"} against schema {"type": "object"} raises TypeError: Field names must be valid identifiers. This is a valid object, so no error should be raised.

This is likely to come up whenever an object is being used as an associative array rather than as a record with a fixed set of fields.

Test case:

from openapi_core import Spec
from openapi_core.testing import MockRequest
from openapi_core.validation.request import openapi_request_validator

spec = Spec.create(
    {
        "openapi": "3.0.3",
        "info": {"title": "test", "version": "0"},
        "paths": {
            "/url": {
                "post": {
                    "parameters": [
                        {
                            "name": "data",
                            "in": "query",
                            "content": {
                                "application/json": {
                                    "schema": {"type": "object"}
                                }
                            },
                        }
                    ],
                    "responses": {"200": {"description": "Success."}},
                }
            }
        },
    },
)
request = MockRequest(
    "http://localhost:9991/", "post", "/url", args={"data": '{"0": "a"}'}
)
result = openapi_request_validator.validate(spec, request)
print(result)
result.raise_for_errors()

[p1c2u]

Hi @andersk

thanks for reporting the issue
enabling dynamic dataclasses by default was bad idea. I will move it to standard dict