How to write a JSON schema for array of objects?

Question

My JSON string would be formatted as:

{
    "count":3,
    "data":[
        {
            "a":{"ax":1}
        },
        {
            "b":{"bx":2}
        },
        {
            "c":{"cx":4}
        }
    ]
}

The data array contains many a and b and c. And no other kinds of objects.

If count==0, data should be an empty array [].

I'm using https://github.com/hoxworth/json-schema to validate such JSON objects in Ruby.

require 'rubygems'
require 'json-schema'

p JSON::Validator.fully_validate('schema.json',"test.json")

The schema.json is:

{
  "type":"object",
  "$schema": "http://json-schema.org/draft-03/schema",
  "required":true,
  "properties":{
     "count": { "type":"number", "id": "count", "required":true },
     "data": { "type":"array", "id": "data", "required":true,
       "items":[
           { "type":"object", "required":false, "properties":{ "a": { "type":"object", "id": "a", "required":true, "properties":{ "ax": { "type":"number", "id": "ax", "required":true } } } } },
           { "type":"object",  "required":false, "properties":{ "b": { "type":"object", "id": "b", "required":true, "properties":{ "bx": { "type":"number", "id": "bx", "required":true } } } } },
           { "type":"object",  "required":false, "properties":{ "c": { "type":"object", "id": "c", "required":true, "properties":{ "cx": { "type":"number", "id": "cx", "required":true } } } } }
       ]
     }
  }
}

But this for test.json will pass the validation while I suppose it should fail:

{
  "count":3,
  "data":[
      {
          "a":{"ax":1}
      },
      {
          "b":{"bx":2}
      },
      {
          "c":{"cx":2}
      },
      {
          "c": {"z":"aa"}
      }
   ]
}

And this as test.json will fail, while I suppose it should pass:

{
  "count":3,
  "data":[
      {
          "a":{"ax":1}
      },
      {
          "b":{"bx":2}
      }
   ]
}

Seems the wrong schema is validating that the data array contains a,b,c once.

What the right schema should be?

Answer 1

From the JSON schema spec, section 5.5. items:

When this attribute value is an array of schemas and the instance
value is an array, each position in the instance array MUST conform
to the schema in the corresponding position for this array. This
called tuple typing.

Your schema definition requires the first three elements of the array to be exactly those 'a', 'b' and 'c' elements. If items is left empty, any array element is allowed. Similarly, if additionalItems is left empty, any additional array element is allowed.

To get what you want, you need to specify "additionalItems": false and for the items, I think the following (somewhat shortened from your definitions) should work:

"items": {
  "type": [
     {"type":"object", "properties": {"a": {"type": "object", "properties": {"ax": { "type":"number"}}}}},
     {"type":"object", "properties": {"b": {"type": "object", "properties": {"bx": { "type":"number"}}}}},
     {"type":"object", "properties": {"c": {"type": "object", "properties": {"cx": { "type":"number"}}}}}
  ]
}