[java] Combining "properties" and "additionalProperties" in one class generates broken Jackson mapping.

[r-alukhanov]

Description

If using both "properties" and "additionalProperties", the generated class extends HashMap and contains plain properties. Like this:

public class Example extends HashMap<...>  {
  @JsonProperty("a")
  private String a;
  ...
}

Such a class cannot be meaningfully deserialised using Jackson. Jackson puts all the fields as key/value pairs into the HashMap ignoring the plain fields. In provided example getA() would return "null", while get("a") contains the value.

openapi-generator version

Used version: openapi-generator-maven-plugin : 3.3.3

OpenAPI declaration file

Example:
  type: object
  properties:
    a:
    type: string
  additionalProperties: true

Suggest a fix/enhancement

I would suggest to generate such a class (which has both "properties" and "additionalProperites") as:

public class Example {
  private String a;
  private HashMap<> otherProperties;
  ...
}

Similar suggestion in "swagger-codegen" project:
swagger-api/swagger-codegen#5187

[cjmamo]

Any updates on this?

[vandeseer]

We are facing the very same issue using openapi-generator-gradle-plugin : 4.0.1 and the jaxrs-jersey generator.

Are there any workarounds beside not using both (i.e. properties and additionalProperties) at the same time? Are there any plans to fix it?

Thanks!

[FatCash]

Also having problems with serialize/deserialize properly because of this.
Desired approach would be to populate a mustache flag like {{hasAdditionalProperties}} on class level (not vars level), similar to {{isMapContainer}} which is currently true if additionalProperties: true
Then we could use it however we want in mustache to add the otherProperties field and necessary json annotations and builders etc..

The automatic extends HashMap<...> should not be there, instead users can add that themselves in mustasche by checking the {{hasAdditionalProperties}} flag.

[wakedeer]

Any updates on this issue?

[harel-e]

I think that having a member of HashMap instead of extending HashMap would address the issue.
@JsonAnyGetter and @JsonAnySetter are designed for this exactly.
Like @FatCash suggested, there could be a feature flag that would allow backward compatibility (though it seems broken right now anyway)

[cgfarmer4]

curious if this was ever fixed?

[B-hamza]

Also having problems with this issue, Any workaround or update ?

[wakedeer]

Also having problems with this issue, Any workaround or update ?

Hey, Unfortunately not! But you can make customization for it. I have made wrapper over gradle plugin and added some logic. I can share it with you if you are interested in it

[cgfarmer4]

@wakedeer I would be interested to see this.

[harel-e]

This is related to issues 6146 and 5515

[benfonty]

Hello, I faced the same problem. I took a look at the code and it happens that if you use the java generator with configOptions.library set to jersey2 (gradle plugin), it works as you expect.

Unfortunately for me I use the spring generator, I don't know why the behavior hasn't been set for all java generators.

I will maybe propose a fix about that.

[ssiegler]

I tried to adapt the fix for spring to the jaxrs-cxf generator. It seems to work fine for our usecase but I am not sure how to prevent this from breaking existing code which may rely on the current behavior. So I am still hesitant to open a pull request.

[jameswynn]

These changes look really good @benfonty . Is there anything left to get this change accepted?

[jameswynn]

@benfonty can you try merging master to your branch and push again? That should solve the build issue.

[benfonty]

@benfonty can you try merging master to your branch and push again? That should solve the build issue.

@jameswynn CI is indeed OK now.

[sirimantraore]

Hello @jameswynn / @benfonty ,

Have you a update for this fix please ?

Thank you so much :)

[jameswynn]

Is there anything blocking this from making it into the 6.0 release? @wing328 I can update the PR again.

[loiccara]

Hi, this issue is blocking my team as well.
@Zomzog , @banlevente , @borsch , @javisst , @cachescrubber , @welshm , @MelleD, @atextor, @manedev79, What is the status on this issue? I tried to contact @jameswynn with no success. I'm happy trying to merge this pull request if need be.

[WeihmannDev]

Hello, is there an issue with the MR for it not to be merged?

[zsait-clearstreet]

Is there a reason this fix is not being merged yet?

[welshm]

#11572 - this likely needs to be updated and revived to tackle this issue.

If anyone wants to create a new pull request based on the content, that would be helpful.

[benfonty]

Hello, I created this pull request a long time ago and did not receive any approval at the time.
If anybody wants to take it over, feel free.

[Mettbrot]

This issue still persists in the Java generator of the latest version. Does anyone have an update on this?