fix: schema aliases can now be dereferenced correctly

[josh-burton]

This PR fixes an issue where schemas that alias a primitive type schema such as string or int cannot be dereferenced correctly.

[josh-burton]

@walsha2 are you able to weigh in here at all?

I think the _checkReferenceTypes needs either a rework or a fix.

The comment on this line seems to suggest it should return true.

[josh-burton]

I have attempted to implement a fix, which works for my test cases.

[walsha2]

@josh-burton I think this is working (generally) as expected. It might make more sense if you take a look at #81 which resolves a somewhat related issue when I was trying to recreate your issue.

In short, the dereference for the field myField is not a String but rather the actual type that is being referenced, MyAlias. So the dereferencing is working as expected and matches the spec.

This package does not resolve (by design) any further than the schema definitions. Since this is a primitive, the way this package resolves it during the code generation step is to explicitly create the typedef:

typedef StringAlias = String;

Then:

  const factory MyObject({
    @JsonKey(includeIfNull: false) StringAlias? myField,
  }) = _MyObject;

See #81. So yea, if you look at it from the lens of the expected generated code from this package, the dereference is correct. If you are looking to further dereference down to the primitive type it might best be done in your own code? Since you are doing your own code generation.

I might also suggest that you take the typedef approach. If an OpenAPI spec defines a type alias, IMHO this is best represented in code generation as a typedef so that the generated code is as close to a 1:1 as the spec making traceability much clearer.

---

Let me know if I misunderstood the issue!

[josh-burton]

@josh-burton I think this is working (generally) as expected. It might make more sense if you take a look at #81 which resolves a somewhat related issue when I was trying to recreate your issue.

In short, the dereference for the field myField is not a String but rather the actual type that is being referenced, MyAlias. So the dereferencing is working as expected and matches the spec.

This package does not resolve (by design) any further than the schema definitions. Since this is a primitive, the way this package resolves it during the code generation step is to explicitly create the typedef:

typedef StringAlias = String;

Then:

  const factory MyObject({
    @JsonKey(includeIfNull: false) StringAlias? myField,
  }) = _MyObject;

See #81. So yea, if you look at it from the lens of the expected generated code from this package, the dereference is correct. If you are looking to further dereference down to the primitive type it might best be done in your own code? Since you are doing your own code generation.

I might also suggest that you take the typedef approach. If an OpenAPI spec defines a type alias, IMHO this is best represented in code generation as a typedef so that the generated code is as close to a 1:1 as the spec making traceability much clearer.

Let me know if I misunderstood the issue!

Thanks for the explanation. I'm not currently using typedefs in my generated code but I will implement it and see if it solves my issue.

Thanks!