Add tests for unevaluatedItems interaction with contains

[marksparkza]

ref #293

[gregsdennis]

I believe there's some discussion around additionalItems and contains where it was decided that contains doesn't affect that evaluation. It's likely that unevaluatedItems follows suit.

I'll see if I can dig that up.

---

Found it: it's the issue that your issue references: json-schema-org/json-schema-spec#810

There's a lot of discussion in there, and it's worth reading.

[marksparkza]

Thanks - I have seen that discussion, and have just read the followup PR: json-schema-org/json-schema-spec#925

I disagree that contains shouldn't affect evaluation. In my view, it is perfectly reasonable to build use cases around the interaction of unevaluatedItems and contains, and I think the 2020-12 spec has it completely right.

Consider the following scenario (I will add this as another test case to this PR). This is a contrived example, but it illustrates how the (non)evaluation of contains may be used to control an item dependency relationship.

An array may contain any sequence of a's, b's and c's, but it may only contain b's if there is at least one a, and it may only contain c's if there is at least one b. Here is the schema:

{
    "if": {
        "contains": {"const": "a"}
    },
    "then": {
        "if": {
            "contains": {"const": "b"}
        },
        "then": {
            "if": {
                "contains": {"const": "c"}
            }
        }
    },
    "unevaluatedItems": false
}

Valid examples:

[]
[ "a", "a" ]
[ "a", "b", "a", "b", "a" ]
[ "c", "a", "c", "c", "b", "a" ]

Invalid examples:

[ "b", "b" ]
[ "c", "c" ]
[ "c", "b", "c", "b", "c" ]
[ "c", "a", "c", "a", "c" ]

[jdesrosiers]

@gregsdennis

there's some discussion around additionalItems and contains where it was decided that contains doesn't affect that evaluation.

Unfortunately, that's not how that ended up. The change got pushed through despite continued strong disagreement.