Skip to content

openapi-32-json-schema-alignment.md

Latest commit

 

History

History
30 lines (23 loc) · 1.93 KB

openapi-32-json-schema-alignment.md

File metadata and controls

30 lines (23 loc) · 1.93 KB
title OpenAPI 3.2: Verify json-schema-faker alignment with draft-bhutton-json-schema-01
parentIssue 1673
labels
enhancement
openapi-3.2
assignees
milestone

OpenAPI 3.2 updates its embedded JSON Schema reference to draft-bhutton-json-schema-01 and draft-bhutton-json-schema-validation-01. The notable additions compared to earlier drafts include a revised unevaluatedProperties, unevaluatedItems, and prefixItems.

Current state

Counterfact uses json-schema-faker to generate random values from schemas. The library's supported JSON Schema draft determines which keywords are available when generating mock data. It is unclear whether the current version of json-schema-faker supports all keywords in draft-bhutton-json-schema-01.

Proposed changes

  • Audit which draft-bhutton-json-schema-01 keywords are used in practice and verify that json-schema-faker handles them correctly
  • In particular, verify support for prefixItems, unevaluatedItems, and unevaluatedProperties
  • If the current version of json-schema-faker does not support these keywords, upgrade the library or add workarounds (e.g. stripping unsupported keywords before passing the schema to the faker)
  • Add tests using schemas that exercise the new keywords to confirm correct random value generation

Acceptance criteria

  • A schema using prefixItems generates random tuple values with correct types for each position
  • A schema using unevaluatedProperties: false does not include extra properties in generated values
  • A schema using unevaluatedItems: false does not include extra items in generated array values
  • All existing random-generation tests continue to pass
  • If json-schema-faker is upgraded, no regressions are introduced in existing mock response generation