Arrays in schemas show up as a separate model

v7.1.0

If I have:

"topten": Joi.array().items(
    Joi.object({
        name: Joi.string().required().label('name').description('user name'),
        score: Joi.number().required().min(0).label('score')
    }).label('TopTen')

I get something like:

{
  topten (topten, optional)
}

topten [
    TopTen
]

TopTen {
    name (string): user name ,
    score (number),
}

I would like to get something like:

{
  topten (Array of TopTen, optional)
}

TopTen {
    name (string): user name ,
    score (number),
}

[moander]

Hi guys,

Is it possible to set an option to avoid each array becoming a separate model? In my case it will reduce the number of models significantly.

Joi.object({
    rows: Joi.array().items(Joi.object({
        name: Joi.string()
    }).label('Row'))
}).label('RowList')

The object above shows up like this in the swagger.json:

"Row": {
    "type": "object",
    "properties": {
        "name": {
            "type": "string"
        }
    }
},
"rows": {
    "type": "array",
    "items": {
        "$ref": "#/definitions/Row"
    }
},
"RowList": {
    "type": "object",
    "properties": {
        "rows": {
            "$ref": "#/definitions/rows"
        }
    }
},

What I want is to reduce it to the following

"Row": {
    "type": "object",
    "properties": {
        "name": {
            "type": "string"
        }
    }
},
"RowList": {
    "type": "object",
    "properties": {
        "rows": {
            "type": "array",
            "items": {
                "$ref": "#/definitions/Row"
            }
        }
    }
},

[irothschild]

Hi Glenn,

First let me thank you for all the hard work you’ve put into this over what now amounts to years. I understand the challenge of keeping things in sync with the fast moving Hapi and Joi targets, let along the swagger and swagger-ui changes.

The real goal for me in using this module is to generate clear, understandable API documentation for end-user developers. Unfortunately, we’re not there yet and in some ways it seems to be moving away from my goal. I don’t claim to understand all the ways people are using this module and for what purposes (code generation seems to be one) but they seem to be different from mine.

For documentation purposes, I don’t think it makes sense to break arrays out as separate objects. It’s just confusing. As well, I don’t understand the use of object definition equality/hashing in the module. For me, if Joi objects have different labels, they are different classes.

Right now if I do:

    a: Joi.object({name: Joi.string().required().label('name')}).required().label('A'),
    b: Joi.object({name: Joi.string().required().label('name')}).required().label('B'),

In the Swagger UI I see:

a (A, optional),
b (A, optional),

A {
  name (string)
}

Apart from them being mislabeled as optional, both are marked as class A when b should be B.

Again, I don’t understand how people are using this module but I think a lot can be simplified in new versions by forcing people to be explicit in their labeling and using those labels without doing any kind of equality tests.

I don’t know where that leaves me. I see three options:

1. Make some changes and do a PR. This might entail more configuration options to avoid breaking things for people who want/need currently functionality. I know you want to avoid adding more options so maybe you’ll reject this idea.
2. Do a “hard-fork”.
3. Generate a swagger.json, either manually editing the hapi-swagger output or using another swagger generation tool.