JSON Schema Validation Issue

[victorgarcia98]

Hello,
Thanks for developing this flexibility standard!

I've been using s2 protocol jsonschema description to validate some entities/messages but I've encountered an issue when validating referenced schemas. Apparently, this tiny example for a FRBC.OperationModeElement passes validation when It shouldn't:

    {
        "fill_level_range": [0.1, 0.2, 0.3],
        "fill_rate":  {
            "start_of_range": 0,
            "end_of_range": 0.1
        },
        "power_ranges": [
            {
                "start_of_range": 5725,
                "end_of_range": 5725,
                "commodity_quantity": "ELECTRIC.POWER.L1"
            }
        ],
        "running_costs": {
            "start_of_range": 0,
            "end_of_range": 0
        }
    }

fill_level_range should be of type NumberRange but passing a list just works. If I use a dict instead of the list, validation works (checks key and values).

I've tried this with python-jsonschema validator and a web validator to check if It was a bug in the implementation of validator but both yield the same result. Not sure if this is an expected behavior of jsonschema or the definition of s2-protocol in jsonschema but I'm not sure how to figure it out so I'm reaching you for some enlightenment.

Thanks a lot in advanced.

Regards,
Víctor

[wilcowijbrandi]

Hi Víctor,

Thanks for reaching out! That JSON message shouldn't successfully validate against the schema. It seems most validators don't follow the reference (e.g. "$ref": "../schemas/NumberRange.schema.json",) without any warnings or errors whatsoever. If you replace the reference with the actual definition the JSON doesn't validate. I'm not sure yet if this is a problem with the mentioned validators or that we do something wrong with defining these references. I'll look into it...

Cheers,
Wilco

[victorgarcia98]

Hi Wilco,

Thank for the response!

I'm not really experienced with jsonschema, but the references look good for me. I think I've tried the same with the asyncapi validator and It didn't validate (as expected).

For now, we're using pydantic models that generated with the tool datamodel-codegen, but we are refactoring the models manually to a better file structure. With pydantic, the inner types are constrained and we get the benefit of being used in frameworks such as FastAPI. So far, we have generated / revised the code for the FRBC operation mode, but we are looking forward to do the others, also in collaboration with other parties using the protocol 😄.

The python module is in here. Feel free to check it out and/or propose any change.

Thanks,
Víctor