JSONMatcherBuilder matcher doesn't handle lists nested inside a sub-tuple

[sbarber2]

From Mic Hussey:

do you know if anyone has implemented handling of matching with nested arrays of tuples?

I discovered that when I wanted to match something like this:

    JSONMatcherBuilder mb = new JSONMatcherBuilder(dequeuerFeatureDataOut.getSchema()).ignoreMissingFields(true);

    String outputTupleAsJSONString = "{\"Entity\":{\"EntityId\":\"ent1\"}, \"DeviceReadings\":["
            +"{\"Device\":{\"PointId\":\"pt1000\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":10.0}]},"
            +"{\"Device\":{\"PointId\":\"pt1001\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":20.0}]},"
            +"{\"Device\":{\"PointId\":\"pt1002\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":30.0}]}"
        +"]}";`

that the matcher stopped recursing once it hit an array.

[sbarber2]

Created issue4 branch to work on this issue. There is a failure in the new testNestsLists() test I wrote, but not sure if this is the same failure that Mic reported. Asked Mic for a failing test to support his issue report.

[sbarber2]

More from Mic on how he would like to use JSON to specify matchers:

[ ... ] "what behaviour do I expect"?

I had the naive expectation that if I set builder.ignoreMissingFields(true) then I could give in a semi-populated tuple tree and the matcher would do exact matches for all fields it found with contents and ignore those it didn't find.

i.e.
if I build the matcher using a tuple like this:

outputTupleAsJSONString = "{\"Entity\":{\"EntityId\":\"ent1\"}, \"DeviceReadings\":["
                +"{\"Device\":{\"PointId\":\"pt1000\",\"ParentEntityId\":\"ent1\"}},"
                +"{\"Device\":{\"PointId\":\"pt1001\",\"ParentEntityId\":\"ent1\"}},"
                +"{\"Device\":{\"PointId\":\"pt1002\",\"ParentEntityId\":\"ent1\"}}"
            +"]}";

it would match this

tupleAsJSONString = "{\"Entity\":{\"EntityId\":\"ent1\"}, \"DeviceReadings\":["
                +"{\"Device\":{\"PointId\":\"pt1000\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":10.0}]},"
                +"{\"Device\":{\"PointId\":\"pt1001\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":20.0}]},"
                +"{\"Device\":{\"PointId\":\"pt1002\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":30.0}]}"
            +"]}";

here there is an extra field called "Readings" which should be ignored.

But this would fail the match (wrong value in DeviceReadings[0].Device.PointId)

tupleAsJSONString = "{\"Entity\":{\"EntityId\":\"ent1\"}, \"DeviceReadings\":["
                +"{\"Device\":{\"PointId\":\"pt2000\",\"ParentEntityId\":\"ent10\"}, \"Readings\":[{\"Value\":10.0}]},"
                +"{\"Device\":{\"PointId\":\"pt1001\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":20.0}]},"
                +"{\"Device\":{\"PointId\":\"pt1002\",\"ParentEntityId\":\"ent1\"}, \"Readings\":[{\"Value\":30.0}]}"
            +"]}";

I didn't really consider what would happen if there was an extra entry in one of the arrays, but we could add a directive matcher.insistExactArrayLength(false) which would determine the behaviour. Similarly I would expect we'd need a matcher.allowOutOfOrderInArray(true) or similar. Here be dragons obviously...