Transform and flatten JSON using GSON

Question

I'm working on splitting a SignalK JSON object into canonical JSON items representing each value.

The original JSON looks like this:

{
"mmsi": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
"name": "Mona",
"navigation": {
    "position": {
        "timestamp": "1991-09-03T03:5:36.000Z",
        "latitude": 51.763691,
        "longitude": 9.501367,
        "altitude": 0.000000,
        "source": "N0183-01"
    },
    "courseOverGroundTrue": {
        "value": 23.000000
    },
    "speedOverGround": {
        "value": 2.010289
    }
},
"environment": {
    "depth": {
        "belowTransducer": {
            "value": 12.700000
        }
    },
    "wind": {
        "angleApparent": {
            "value": 0.174533
        },
        "speedApparent": {
            "value": 0.000000
        }
    }}}

The needed transformed JSON looks like this, with JSON elements representing each value, and with item naming representing the whole path of the value.

{
"items": [{
        "columns": {
            "assetId": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
            "description": "EnvironmentWindSpeedApparent",
            "isStep": false,
            "name": "EnvironmentWindSpeedApparent",
            "timestamps": 1523962903470,
            "type": "numerical",
            "values": 0.0
        },
        "key": "20180417-130143470EnvironmentWindSpeedApparent5377770-4ee4-4a4b-3230-888037332031"
    }, {
        "columns": {
            "assetId": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
            "description": "EnvironmentWindAngleApparent",
            "isStep": false,
            "name": "EnvironmentWindAngleApparent",
            "timestamps": 1523962903470,
            "type": "numerical",
            "values": 0.174533
        },
        "key": "20180417-130143470EnvironmentWindAngleApparent5377770-4ee4-4a4b-3230-888037332031"
    }, {
        "columns": {
            "assetId": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
            "description": "EnvironmentDepthBelowTransducer",
            "isStep": false,
            "name": "EnvironmentDepthBelowTransducer",
            "timestamps": 1523962903470,
            "type": "numerical",
            "values": 12.7
        },
        "key": "20180417-130143470EnvironmentDepthBelowTransducer5377770-4ee4-4a4b-3230-888037332031"
    }, {
        "columns": {
            "assetId": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
            "description": "NavigationPositionLongitude",
            "isStep": false,
            "name": "NavigationPositionLongitude",
            "timestamps": 1523962903470,
            "type": "numerical",
            "values": 9.501367
        },
        "key": "20180417-130143470NavigationPositionLongitude5377770-4ee4-4a4b-3230-888037332031"
    }, {
        "columns": {
            "assetId": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
            "description": "NavigationPositionLatitude",
            "isStep": false,
            "name": "NavigationPositionLatitude",
            "timestamps": 1523962903470,
            "type": "numerical",
            "values": 51.763691
        },
        "key": "20180417-130143470NavigationPositionLatitude5377770-4ee4-4a4b-3230-888037332031"
    }, {
        "columns": {
            "assetId": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
            "description": "NavigationCourseOverGroundTrue",
            "isStep": false,
            "name": "NavigationCourseOverGroundTrue",
            "timestamps": 1523962903470,
            "type": "numerical",
            "values": 23.0
        },
        "key": "20180417-130143470NavigationCourseOverGroundTrue5377770-4ee4-4a4b-3230-888037332031"
    }, {
        "columns": {
            "assetId": "urn:mrn:signalk:uuid:5377770-4ee4-4a4b-3230-888037332031",
            "description": "NavigationSpeedOverGround",
            "isStep": false,
            "name": "NavigationSpeedOverGround",
            "timestamps": 1523962903470,
            "type": "numerical",
            "values": 2.010289
        },
        "key": "20180417-130143470NavigationSpeedOverGround5377770-4ee4-4a4b-3230-888037332031"
    }
    ]}

How to do this transformation in a flexible way that adopts to changing sub-nodes being available in the original JSON ? I'm transforming it now in a simplistic way, but would like to know if it can be done using JsonReader , gson or other ways of iterating through the original JSON object.

Answer 1

I ended up defining the object structure representing the transformed json, and writing a custom serializer to do the transformation. It could be more efficient ways of doing it, but this approach seems to be working OK.

public class SignalKDeserializer implements JsonDeserializer<TargetObject> {
//written based on examples from http://www.javacreed.com/gson-deserialiser-example/

final TargetObject targetObject = new TargetObject ();

@Override
public TargetObject deserialize(final JsonElement json, final Type typeOfT, final JsonDeserializationContext context)
        throws JsonParseException {
    final JsonObject jsonObject = json.getAsJsonObject();
    traverse (jsonObject,0,"", mmsi);
    return targetObject;

}

private void traverse (JsonObject jsonObject, Integer level, String parentName, String mmsi) {
    Set<String> keys = jsonObject.keySet();
    Iterator<?> keysIterator = keys.iterator ();
    while( keysIterator.hasNext ()) {
        String key = (String) keysIterator.next ();
        String signalName = parentName+upperCaseFirst (key); //setting signalName to complete path of value
        if (jsonObject.get (key) instanceof JsonObject) {
            traverse ((jsonObject.get (key)).getAsJsonObject (),level+1,upperCaseFirst (signalName),mmsi);
        } else if (jsonObject.get (key) instanceof JsonElement) {
            if (level>0) {
                try {
                    final Double value = jsonObject.get(key).getAsDouble ();
                    calendar = Calendar.getInstance ();
                    Long timeStamp = calendar.getTimeInMillis ();
                    Item targetItem = new Item ();
                    targetItem.columns.setIsStep (false);
                    targetItem.columns.setAssetId (mmsi);
                    targetItem.columns.setTimestamps (calendar.getTimeInMillis ());
                    targetItem.columns.setType ("numerical");
                    targetItem.columns.setDescription (signalName);
                    targetItem.columns.setValues (value);
                    targetItem.columns.setName (signalName);
                    targetItem.setKey (signaldateformat.format (timeStamp) + mmsi);
                    targetObject.items.add (targetItem);
                }
                catch (NumberFormatException n) {
                    // Expected, the value is non numerical and  will not be transformed
                }
            }
        }
    }
}}

I'm using the serializer from my main class like this:

final GsonBuilder gsonBuilder = new GsonBuilder();
gsonBuilder.registerTypeAdapter(TargetObject.class, new SignalKDeserializer ());
final Gson gson = gsonBuilder.create();
TargetObject targetObject = gson.fromJson (jsonSignalK,TargetObject.class);
String jsonOutString = gson.toJson (targetObject);

jsonOutString contains the transformed json I needed.