refactor: simplify type handling in ModelDeserializer

[Mattias-Sehlstedt]

Pull Request

Thank you for contributing to swagger-core!

Please fill out the following information to help us review your PR efficiently.

---

Description

Simplifies the logic in the ModelDeserializer so that a typed Schema is clearly handled as its own separate scenario with its own processing. Also reduced the surface area against Jackson, so that an upgrade will have a much smaller delta (this change was mainly inspired by this larger-than-necessary delta).

Type of Change

• 🐛 Bug fix
• ✨ New feature
• ♻️ Refactor (non-breaking change)
• 🧪 Tests
• 📝 Documentation
• 🧹 Chore (build or tooling)

Checklist

• I have added/updated tests as needed
• I have added/updated documentation where applicable
• The PR title is descriptive
• The code builds and passes tests locally
• I have linked related issues (if any)

Screenshots / Additional Context

[Rymert-Mostert]

My bad. Just looking for the "Dismiss my own review" button.
https://github.com/orgs/community/discussions/120006

[ewaostrowska]

Hi @Mattias-Sehlstedt!

Thank you for contributing to Swagger Core again and for working to improve the code.

Overall, the refactor looks good, but I found one place where the behavior changes. I’ve pushed tests that illustrate the difference.

Please decide whether $ref should continue using asText() or require a textual node. If strict textual validation is intentional, please document it as a behavioral change

[Mattias-Sehlstedt]

Good spotting. I cannot strongly prove that the changes will not affect someone else's workflow, so it is safest to ensure that the behavior is precisely as before.

I have re-added the TextNode casting so that the entire type-branch will throw if there is a casting issue (same behavior as before since array is the first if-case).

I have also restored how $ref is read (with .asText()) and also added a test to showcase what that results in (.textValue() resulted in null as the String value).