Adds fix for composed schema model vars

[spacether]

Adds fix for composed schema model vars
Per guidance from json schema and openapi people, when a composed schema has properties or additionalProperties
object payloads are validated to the properties or additionalProerties definition at the composed schema level, independent of properties in composed oneOf/anyOf/allOf definition

So that means that CodegenModel.vars must store only the properties present in the composed schema, if they are present.
This has implications to how we validate composed schemas across generators, and that all generators should eventually support supportsAdditionalPropertiesWithComposedSchema = True.
For now only python and java support it.

This is a fix for: #8813

Blocker

Landing this PR is blocked by: #8802
because we need additionalProperties to be set at the model level before we can use it to allow any value in in all key value pairs.

@sebastien-rosset should the go generator set supportsAdditionalPropertiesWithComposedSchema to True?

PR checklist

• Read the contribution guidelines.
• Pull Request title clearly describes the work in the pull request and Pull Request description provides details about how to validate the work. Missing information here may result in delayed response from the community.
• Run the following to build the project and update samples:

  ./mvnw clean package 
  ./bin/generate-samples.sh
  ./bin/utils/export_docs_generators.sh
  

  Commit all changed files.
  This is important, as CI jobs will verify all generator outputs of your HEAD commit as it would merge with master.
  These must match the expectations made by your contribution.
  You may regenerate an individual generator by passing the relevant config(s) as an argument to the script, for example ./bin/generate-samples.sh bin/configs/java*.
  For Windows users, please run the script in Git BASH.
• File the PR against the correct branch: master, 5.1.x, 6.0.x
• If your PR is targeting a particular programming language, @mention the technical committee members, so they are more likely to review the pull request.

[wing328]

Per guidance from json schema and openapi people, when a composed schema has properties or additionalProperties

Can you please share the links to the discussions?

[spacether]

Per guidance from json schema and openapi people, when a composed schema has properties or additionalProperties

Can you please share the links to the discussions?

Sure, here it is: OAI/OpenAPI-Specification#2478

[wing328]

@spacether the declawed property has been removed as part of the change, which doesn't seem to be correct according to the definition of Cat:

    Cat:
      allOf:
        - $ref: '#/components/schemas/Animal'
        - $ref: '#/components/schemas/Address'
        - type: object
          properties:
            declawed:
              type: boolean

ref: https://github.com/OpenAPITools/openapi-generator/blob/master/modules/openapi-generator/src/test/resources/3_0/java/petstore-with-fake-endpoints-models-for-testing-with-http-signature.yaml#L1411-L1418

[spacether]

The CodegenModel Cat not containing declawed when vars only contains properties in the current schema is correct. declawed definition only exists in that allOf third schema. It does not have an explicit definition inside Cat because cat has by default:

Cat:
  properties: {} # no declawed property here
  additionalProperties: true # this allows ingestion of any type of value

Cat containing allOf properties in generated code is an implementation decision.
Per our chat, an allOf schema could be composed also, which means that the variables that it includes in a composed schema like Cat can change from one payload to another or could event be different types like null or object.
This is why I think putting this behind a CLI option may be the best path forward.
Fixing Java looks non-trivial to me but maybe I am missing something.

Do you have a straightforward way to fix Java in this PR?
How do you think I should proceed here?

[wing328]

The CodegenModel Cat not containing declawed when vars only contains properties in the current schema is correct

Can you share the link(s) to discussion supporting this claim so that everyone is on the same page?

[spacether]

The point of this PR is that vars should only contain the properties defined in properties of the current scheme. It should not include composed schema properties. This is consistent with how the openapi folks told me to interpret schemas.

Yup it is in the link I pasted above OAI/OpenAPI-Specification#2478

And here is the slack discussion https://json-schema.slack.com/archives/C5CF75URH/p1613937923034800

CodegenModel is one of our classes which implemente Schema and includes some extra info.

[spacether]

So this change works to remove the vars that were hoisted into composed schemas from those composed schemas. They still exist in the source schemas and can be assigned as additional properties in the composed schemas.
One would still need to add the composed schema propertyName property back in to schemas that use it.

I am concerned about merging this because it will remove the variables from the properties and signatures of the composed schemas. While not a breaking change, I suspect that many users will not want that.
It may make sense for me to segregate this feature into my future python-experimental generator.