Using JSON schema with an array of mixed object types about json-schema-validator

[fge]

This is a valid schema for draft v3, yes, but not draft v4. And draft v4 is the default.

You have two choices:

• add "$schema": "http://json-schema.org/draft-03/schema#" at the root of your schema, since the library detected its presence and switches versions automatically;
• if you want to use draft v3 as the default, you have to customize your ValidationConfiguration:

final ValidationConfiguration cfg = ValidationConfiguration.newBuilder()
    .setDefaultVersion(SchemaVersion.DRAFTV3).freeze();
final JsonSchemaFactory factory = JsonSchemaFactory.newBuilder()
    .setValidationConfiguration(cfg).freeze();

Or of course, you can use a draft v4 schema:

{
    "type" : "array",
    "items" : { 
        "anyOf" : [ 
            { "type" : "object", "properties" : { "name" : { "type" : "string" }}},
            { "type" : "object", "properties" : { "price" : { "type" : "number" } } }
        ]
    }
}

[fge]

Please note that it is not because you bind array elements to a certain schema that these elements must be there. If, for instance, you want your array to have at least these two elements then you must use minItems. Similarly, if you want your array to have a limited number of elements, use maxItems.

Here is for instance a schema constraining an array to have two items, the first being an integer and the second a boolean:

{
    "type": "array",
    "items": [ { "type": "integer" }, { "type": "boolean" } ],
    "minItems": 2,
    "maxItems": 2
}

[fge]

Ah, you didn't mention that you wanted schemas at specific indices in the array. In that case, your schema is not doing it correctly. Your schema should be this:

{
    "type": "array",
    "items": [
        { "schema": [ "for", "first", "element", "here" ] },
        { "schema": [ "for", "second", "element", "here" ] }
    ],
    "etc": "etc"
}

In your other example, the array would not be limited in the number of items but all items would share the same schema.

[ettig]

Thanks!
one thing though - I'm not sure the anyOf will help me with the restriction i'm trying to achieve:
I want to restrict the array to hold couples of [country, region](it's not in the json schema example I wrote above but it's what i need to perform in my schema)
I want the country to be the first element in the array and to be selected from an enum of countries,
the region to be the second element in the array and to be selected from an enum of regions.

I tried the following schema hoping this will do the trick:
{
"type": "array",
"items": {
"type":[
{"type":"string","enum":["FR","BE"]},
{"type":"string","enum":["EU","US"]}
]
},
"minItems": 1,
"uniqueItems": true,
"maxItems": 2
}

the question is if the anyOf is suitable...

[ettig]

Thanks!
sorry I wasn't clear..
I'll give it a try, thanks a lot!

[ettig]

and this can work with draft v4 ?

So I should write it like this?:
{
"type": "array",
"items": [
{"type":"string","enum":["FR","BE"]},
{"type":"string","enum":["EU","US"]}
],
"minItems": 1,
"maxItems": 2
}

[fge]

The items keyword hasn't changed semantics with draft v4, so yes, this will work.

Note that you don't need "type": "string" since an instance is valid against enum if it is exactly equal to one of the elements in the enum. Therefore, enum is enough here.

[ettig]

thank you so much for your quick and helpful replies,
I'll try this :)

[ettig]

works great thanks!

[fge]

Good to hear ;) Can I close the issue?

[ettig]

Yes, thanks again for your help :)

[fge]

OK, thanks! Do not hesitate if you have further questions!

Note: for pure schema-writing related problems, you may join the Google group, or channel #jsonschema on Freenode.

[kof]

shouldn't items contain schema or array of schemas? if so using anyOf should not be inside of items, but instead of items ...

[fge]

@kof sorry, I don't quite make up your question here. What do you have in mind? Remember that if the value of items is an array, each element of this array is the schema the instance element must obey at that particular index.

[agapare]

Hi guys, I need help build a json schema from the sample below:

{
"1453235400": {
"ERRORS_OUT": null,
"OCTETS_OUT": 232.957375822694,
"PACKETS_OUT": 1.8996969225584,
"DISCARDS_OUT": null,
"SPEED": 2048000,
"DISCARDS_IN": null,
"PACKETS_IN": 1.86311703463372,
"OCTETS_IN": 238.160971788575,
"ERRORS_IN": null,
"IFSTATUS": 1
},
"1453235700": {
"ERRORS_OUT": null,
"OCTETS_OUT": 231.133641048357,
"PACKETS_OUT": 1.88395658914729,
"DISCARDS_OUT": null,
"SPEED": 2048000,
"DISCARDS_IN": null,
"PACKETS_IN": 1.84730778885197,
"OCTETS_IN": 236.086393503138,
"ERRORS_IN": null,
"IFSTATUS": 1
},
"1453236000": {
"ERRORS_OUT": null,
"OCTETS_OUT": 230.8488,
"PACKETS_OUT": 1.88284444444444,
"DISCARDS_OUT": null,
"SPEED": 2048000,
"DISCARDS_IN": null,
"PACKETS_IN": 1.84617777777778,
"OCTETS_IN": 235.842177777778,
"ERRORS_IN": null,
"IFSTATUS": 1
}
}

[huggsboson]

@agapare You should start a new issue for this... it should be pretty doable but I'm going to need 20 free minutes to work it out.

[tigerqiu712]

this issue was fixed ?
{
"1453235400": {
"ERRORS_OUT": null,
"OCTETS_OUT": 232.957375822694,
"PACKETS_OUT": 1.8996969225584,
"DISCARDS_OUT": null,
"SPEED": 2048000,
"DISCARDS_IN": null,
"PACKETS_IN": 1.86311703463372,
"OCTETS_IN": 238.160971788575,
"ERRORS_IN": null,
"IFSTATUS": 1
},
"1453235700": {
"ERRORS_OUT": null,
"OCTETS_OUT": 231.133641048357,
"PACKETS_OUT": 1.88395658914729,
"DISCARDS_OUT": null,
"SPEED": 2048000,
"DISCARDS_IN": null,
"PACKETS_IN": 1.84730778885197,
"OCTETS_IN": 236.086393503138,
"ERRORS_IN": null,
"IFSTATUS": 1
},
"1453236000": {
"ERRORS_OUT": null,
"OCTETS_OUT": 230.8488,
"PACKETS_OUT": 1.88284444444444,
"DISCARDS_OUT": null,
"SPEED": 2048000,
"DISCARDS_IN": null,
"PACKETS_IN": 1.84617777777778,
"OCTETS_IN": 235.842177777778,
"ERRORS_IN": null,
"IFSTATUS": 1
}
}