Correct way of using isinstance with this library?

[jtreminio]

Seems depending on the OAS version used, a different set of libraries will be loaded v3_0 vs v3_1.

This is fine, and a great help, but I seem to be running into issues when testing an object with isinstance.

With the following simple OAS

openapi: 3.1.0
info:
  version: 1.0.0
  title: Fixture - Single requestBody
tags:
  - name: pet
paths:
  /single_request_body:
    post:
      tags:
        - pet
      operationId: single_request_body
      responses:
        '200':
          description: Successful operation
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Pet'
  /multiple_request_body:
    post:
      tags:
        - pet
      operationId: multiple_request_body
      responses:
        '200':
          description: Successful operation
      requestBody:
        content:
          application/x-www-form-urlencoded:
            schema:
              $ref: '#/components/schemas/Customer'
          application/json:
            schema:
              $ref: '#/components/schemas/Pet'
components:
  schemas:
    Customer:
      properties:
        id:
          type: integer
          example: 100000
      type: object
    Pet:
      properties:
        id:
          type: integer
          example: 10
      type: object

and the following code:

import openapi_pydantic as oa

# define path to oas_file 
oas_file = ""

oa_data = oa.parse_obj(oas_file)
customer = oa_data.components.schemas.get("Customer")
pet = customer.properties.get("pet")

if isinstance(pet, oa.Reference):
    print(1)
else:
    print(0)

When openapi is 3.1.0 the isinstance check prints 1, but when it is 3.0.2 then we get 0.

Is there a better way to handle this, without knowing ahead of time the version of the OAS to be loaded?

Right now I've fallen back to using if hasattr(pet, "ref") when I want to check if the current object is a Reference, but this doesn't scale very well.

Suggestions?

[mike-oakley]

Hi @jtreminio - you're right that the top-level imports default to the latest version (3.1.x) objects for convenience. However both the 3.0 and 3.1 objects can also be imported explicitly from openapi_pydantic.v3.v3_0 and openapi_pydantic.v3.v3_1 too.

You should therefore either be able to use the isinstance check against the explicit version of Reference that you are expecting (in this case from openapi_pydantic.v3.v3_0 import Reference), or you can import both the v3_0 and v3_1 models and use the tuple form to check both models (ie isinstance(obj, (v3_0.Reference, v3_1.Reference))).

Hope that helps!

[jtreminio]

Thank you, that's what I figured would be the case.

Love this project, thank you for the amazing work.