RepositoryConstraintViolationExceptionMessage produces inaccurate output [DATAREST-805]

[spring-projects-issues]

Christian Bongiorno opened DATAREST-805 and commented

The invalidValue field of RepositoryConstraintViolationExceptionMessage.ValidationError is a string even if the failed member is of another type because it is being string formatted. By maintaining the original type we allow better debugging by expressing more closely the true type input and not just its toString() and also allow the original marshaller/unmarshaller to express the object as is more appropriate.

Example input:

{
    "givenName" : "null",
    "familyName" : "Bongiorno",
    "dob" : "2007-12-03",
    "nationalId" : "12345687",
    "photo" : null
    
}

or

{
    "givenName" : null,
    "familyName" : "Bongiorno",
    "dob" : "2007-12-03",
    "nationalId" : "12345687",
    "photo" : null
    
}

example output

{
    "errors": [
        {
            "entity": "Worker",
            "message": "may not be null",
            "invalidValue": "null",
            "property": "givenName"
        }
    ]
}

Entity:

@Entity
public class Worker {

    @Id
    private Long id;

    @NotNull
    @Pattern(regexp = "[^null]", message = "may not be null")
    private String givenName;

    @NotNull
    private String familyName;

    private LocalDate dob;

    private String nationalId;

    private byte[] photo;
}

I intentionally added a silly validation constraint to demonstrate

---

Affects: 2.4.4 (Gosling SR4), 2.5.1 (Hopper SR1)

Referenced from: commits 29f90e0, c8abdf0

Backported to: 2.5.2 (Hopper SR2)

[spring-projects-issues]

Oliver Drotbohm commented

Thanks for filing that one. As I understand this correctly is that you'd expect a plain null instead of "null", right

[spring-projects-issues]

Christian Bongiorno commented

Yes. And, if it were and out-of-range integer I would expect an unquoted json integer.

I already made the changes locally.