Why is JSON-validation always successful using this schema containing 'allOf'?

Question

I have a JSON schema with which I want to validate some data, using python and the jsonschema module. However, this doesn't quite work as expected, as some of the accepted data doesn't appear valid at all (to me and the purpose of my application). Sadly, the schema is provided, so I can't change the schema itself - at least not manually.

This is a shortened version of the schema ('schema.json' in code below):

{
  "type": "object",
  "allOf": [
    {
      "type": "object",
      "allOf": [
        {
          "type": "object",
          "properties": {
            "firstName": {
              "type": "string"
            },
            "lastName": {
              "type": "string"
            }
          }
        },
        {
          "type": "object",
          "properties": {
            "language": {
              "type": "integer"
            }
          }
        }
      ]
    },
    {
      "type": "object",
      "properties": {
        "addressArray": {
          "type": "array",
          "items": {
            "type": "object",
            "properties": {
              "streetNumber": {
                "type": "string"
              },
              "street": {
                "type": "string"
              },
              "city": {
                "type": "string"
              }
            }
          }
        }
      }
    }
  ]
}

This is an example of what should be a valid instance ('person.json' in code below):

{
  "firstName": "Sherlock",
  "lastName": "Holmes",
  "language": 1,
  "addresses": [
    {
      "streetNumber": "221B",
      "street": "Baker Street",
      "city": "London"
    }
  ]
}

This is an example of what should be considered invalid ('no_person.json' in code below):

{
  "name": "eggs",
  "colour": "white"
}

And this is the code I used for validating:

from json import load
from jsonschema import Draft7Validator, exceptions

with open('schema.json') as f:
    schema = load(f)
with open('person.json') as f:
    person = load(f)
with open('no_person.json') as f:
    no_person = load(f)

validator = Draft7Validator(schema)
try:
    validator.validate(person)
    print("person.json is valid")
except exceptions.ValidationError:
    print("person.json is invalid")
try:
    validator.validate(no_person)
    print("no_person.json is valid")
except exceptions.ValidationError:
    print("no_person.json is invalid")

Result:
person.json is valid
no_person.json is valid

I expected no_person.json to be invalid. What can there be done to have only data such as person.json to be validated successfully? Thank you very much for your help, I'm very new to this (spent ages searching for an answer).

Answer 1

This is work schema and pay attention on "required" (when there is no such key - if field is doesn't get it just skipped):

{
  "type": "object",
  "properties": {
      "firstName": {
        "type": "string"
      },
      "lastName": {
        "type": "string"
      },
      "language": {
        "type": "integer"
      },
      "addresses": {
        "type": "array",
        "items": {
            "type": "object",
            "properties": {
              "streetNumber": {
                "type": "string"
              },
              "street": {
                "type": "string"
              },
              "city": {
                "type": "string"
              }
            },
            "required": [
              "streetNumber",
              "street",
              "city"
           ]
        }
      }
  },
  "required": [
    "firstName",
    "lastName",
    "language",
    "addresses"
  ]
}

I've got:

person.json is valid
no_person.json is invalid

If you have hardest structure of response (array of objects, which contain objects etc) let me known