Inconsistent OpenAPI schema naming with SNAKE_CASE: some Java record fields remain camelCase

[AgilAghamirzayev]

Title

Inconsistent OpenAPI schema naming with SNAKE_CASE: some Java record fields remain camelCase

Describe the bug

When spring.jackson.property-naming-strategy: SNAKE_CASE is enabled, OpenAPI schema generation is inconsistent for Java record fields.
Most fields are converted to snake_case, but some fields remain camelCase in /v3/api-docs and Swagger UI.

In my case, non-boolean fields starting with is... syllable (e.g. issuanceDate, issuingCountry, issuingAuthority) stay camelCase, while other fields are snake_case.

---

To Reproduce

Steps to reproduce the behavior:

• Spring Boot version: 4.0.5
• springdoc-openapi module/version: org.springdoc:springdoc-openapi-starter-webmvc-ui:3.0.2
• Related transitive swagger-core version: io.swagger.core.v3:swagger-core-jakarta:2.2.43

application.yaml:

spring:
  jackson:
    property-naming-strategy: SNAKE_CASE

Optional OpenAPI config used:

@Bean
public ModelResolver modelResolver() {
    return new ModelResolver(
        Json.mapper().copy().setPropertyNamingStrategy(PropertyNamingStrategies.SNAKE_CASE)
    );
}

Sample DTO:

public record PidLookupResponse(
    String familyName,
    String expiryDate,
    String issuanceDate,
    String issuingCountry,
    String issuingAuthority
) {}

Sample controller:

@RestController
@RequestMapping("/api/test")
public class HelloController {

    @GetMapping
    public PidLookupResponse get() {
        return new PidLookupResponse(
            "Doe", "2030-01-01", "2020-01-01", "AZ", "Authority"
        );
    }
}

Run app and check:

• GET /v3/api-docs
• components.schemas.PidLookupResponse.properties

Actual result (OpenAPI JSON):

{
  "family_name": { "type": "string" },
  "expiry_date": { "type": "string" },
  "issuanceDate": { "type": "string" },
  "issuingCountry": { "type": "string" },
  "issuingAuthority": { "type": "string" }
}

Expected result (OpenAPI JSON):

{
  "family_name": { "type": "string" },
  "expiry_date": { "type": "string" },
  "issuance_date": { "type": "string" },
  "issuing_country": { "type": "string" },
  "issuing_authority": { "type": "string" }
}

---

Expected behavior

All schema property names should consistently follow SNAKE_CASE when global Jackson naming strategy is set to SNAKE_CASE.

Screenshots

Can be provided if needed (Swagger UI shows mixed snake_case + camelCase in the same schema).

Additional context

Runtime JSON response naming is snake_case as expected.
Issue is specifically about generated OpenAPI schema naming.
Workaround is adding explicit @JsonProperty(...) on affected fields, but this should not be required for global naming strategy.

[Mattias-Sehlstedt]

Hi, this issue is entirely tied to the swagger-core library that does the model introspection. Can be verified with:

ModelResolver modelResolver = new ModelResolver(
        Json.mapper().copy().setPropertyNamingStrategy(PropertyNamingStrategies.SNAKE_CASE)
);
ModelConverterContextImpl context = new ModelConverterContextImpl(modelResolver);
Schema<?> schema = modelResolver.resolve(new AnnotatedType(PidLookupResponse.class), context, null);

And then you will see that the field names vary.