Get value from Embedded Document Mongo Java

Question

I have the following document in mongo:

>  {    "_id": ObjectId("569afce4b932c542500143ec"),    
>    "date": "2016-1-17T2:31:0Z",    
>    "day": NumberInt(17),    
>    "model1": {
>      "date": "2016-01-17T02:31+0000",
>      "MondayModel": {
>        "gtxdotdot": {
>          "xdotdot": 0,
>          "xdot": 0
>       },
>        "lsxdotdot": {
>          "xdotdot": 0,
>          "xdot": 0
>       },
>        "gtxdot": {
>          "xdotdot": 0,
>          "xdot": 0
>       },
>        "lsxdot": {
>          "xdotdot": 0,
>          "xdot": 0
>       },
>        "modeldotdot": {
>          "mean": 0,
>          "sdvar": 0
>       },
>        "modeldot": {
>          "mean": 0,
>          "sdvar": 0
>       }
>     } 
>     }

I wish to both find this document and extract only the values of model1.MondayModel.gtxdotdot.xdotdot/xdot/mean/sdvar ...

My current code does so with the following:

MongoCursor<Document>  back = collection.find(and(eq("topic",topic),eq("sp",sp))).limit(1).iterator();
if (back.hasNext())
{
    Document doc = back.next();
    Document tmpddc1 = (Document)doc.get("model1");
    Document tmpddc2 = (Document)tmpddc1.get("MondayModel");

    Document tmpddc3 = (Document)tmpddc2.get("gtxdotdot");
    gtxdotdotXdotdot = tmpddc3.getDouble("xdotdot");
    gtxdotdotXdot    = tmpddc3.getDouble("xdot");

             tmpddc3 = (Document)tmpddc2.get("lsxdotdot");
    lsxdotdotXdotdot = tmpddc3.getDouble("xdotdot");
    lsxdotdotXdot    = tmpddc3.getDouble("xdot");

             tmpddc3 = (Document)tmpddc2.get("gtxdot");
    gtxdotXdotdot = tmpddc3.getDouble("xdotdot");
    gtxdotXdot    = tmpddc3.getDouble("xdot");            

             tmpddc3 = (Document)tmpddc2.get("lsxdot");
    lsxdotXdotdot = tmpddc3.getDouble("xdotdot");
    lsxdotXdot    = tmpddc3.getDouble("xdot");  


             tmpddc3 = (Document)tmpddc2.get("modeldotdot");
    modeldotdotXmean = tmpddc3.getDouble("mean");
    modeldotdotXsdvar    = tmpddc3.getDouble("sdvar");             

             tmpddc3 = (Document)tmpddc2.get("modeldot");
    modeldotXmean = tmpddc3.getDouble("mean");
    modeldotXsdvar    = tmpddc3.getDouble("sdvar");                         

}

Instead of running thought he document (as above) is there a way to get the values using the dot notation [model1.MondayModel.gtxdotdot.xdotdot] ? Something such as:

double value =  doc.getDouble("model1.MondayModel.gtxdotdot.xdotdot");

Argentiferous

Answer 1

I don't think you can use dot notation directly, but you can create your own helper function.

Solution 1: get field with dot notation

public static Object getWithDotNotation( Document document, String dots ) 
     throws MongoException{

    String[] keys = dots.split( "\\." );
    Document doc = document;

    for( int i = 0; i < keys.length - 1; i++ ){
        Object o = doc.get( keys[ i ] );
        if( o == null || !( o instanceof Document ) ){
            throw new MongoException( String.format( 
                    "Field '%s' does not exist or s not a Document", keys[ i ] ) );
        }
        doc = ( Document ) o;
    }//end for

    return doc.get( keys[ keys.length - 1 ] );
}

You can then use it like this:

String dotNotation = "model1.MondayModel.gtxdotdot.xdotdot";

FindIterable<Document> projection = mongoColl.find()
    .projection( fields( include( dotNotation ) ) );

Double value = ( Double ) getWithDotNotation( projection.first(), dotNotation );

System.out.println( value ); // result: 0.0

This will simplify your code a whole lot. The only things to care for are:

• if you are not sure of you dot notation, use a try catch block
• the method getWithDotNotation might return null
• be careful about forced casts, use it only if you are 100% sure of the type of your data (here Double).

Solution 2: flatten your doc

public static Document flattenDoc( Document document ){

    Document flattened = new Document();
    Queue<Pair<String, Document>> queue = new ArrayDeque<>();
    queue.add( new Pair<>( "", document ) );

    while( !queue.isEmpty() ){
        Pair<String, Document> pair = queue.poll();
        String key = pair.getKey();
        for( Map.Entry<String, Object> entry : pair.getValue().entrySet() ){
            if( entry.getValue() instanceof Document ){
                queue.add( new Pair<>( key + entry.getKey() + ".", ( Document ) entry.getValue() ) );

            }else{
                flattened.put( key + entry.getKey(), entry.getValue() );

            }
        }//end for
    }

    return flattened;
}

With your sample data, the result of flattenDoc is the following:

Document{{_id=569afce4b932c542500143ec,
date=2016-1-17T2:31:0Z,
day=17,
model1.date=2016-01-17T02:31+0000,
model1.MondayModel.gtxdotdot.xdotdot=0.0,
model1.MondayModel.gtxdotdot.xdot=0.0,
model1.MondayModel.lsxdotdot.xdotdot=0.0,
model1.MondayModel.lsxdotdot.xdot=0.0,
model1.MondayModel.gtxdot.xdotdot=0.0,
model1.MondayModel.gtxdot.xdot=0.0,
model1.MondayModel.lsxdot.xdotdot=0.0,
model1.MondayModel.lsxdot.xdot=0.0,
model1.MondayModel.modeldotdot.mean=0.0,
model1.MondayModel.modeldotdot.sdvar=0.0,
model1.MondayModel.modeldot.mean=0.0,
model1.MondayModel.modeldot.sdvar=0.0}}

So you can use getDouble("model1.MondayModel.gtxdotdot.xdotdot") directly. This approach might be more efficient if you need to access all the fields.

Cymar

Shouldn't this be for(int i = 0; i < keys.length; i++) instead? – Desjardins

no, because the last value is the name of a document property, not a document. So you need to stop at the name before the last dot (hope this explanation is clear?) – Cymar

Answer 2

You can do it one of three ways.

You can use aggregation framework to project the value of embedded field using dot notation.

Using Aggregation

 import static com.mongodb.client.model.Aggregates.*;
 import static com.mongodb.client.model.Filters.eq;
 import static com.mongodb.client.model.Projections.computed;
 import static java.util.Arrays.*;
 import static com.mongodb.client.model.Projections.include;

 MongoClient mc = new MongoClient();
 MongoDatabase db = mc.getDatabase("test");
 MongoCollection<Document> collection = db.getCollection("collection");
 Document document = 
     collection.aggregate(asList(
        match(eq("day",17)),
        project(computed("val", "$model1.MondayModel.gtxdotdot.xdotdot")))).
     first();
 Double embeddedField = document.getDouble("val");

Using Distinct

 Double embeddedField = collection.distinct("model1.MondayModel.gtxdotdot.xdotdot", eq("day",17), Double.class).first();

Using Find

 Document document = collection.find(eq("day",17)).projection(include("model1.MondayModel.gtxdotdot.xdotdot")).first();
 Double embeddedField = document.get("model1", Document.class).get("MondayModel", Document.class).get("gtxdotdot", Document.class).getDouble("xdotdot")

Shelleyshellfire

Your first two examples refer to querying rather getting values from Document objects. However, your second line of your 'Using Find' example is valid. – Desjardins

Getting values directly from Document is not possible. Refer here. As a alternative I have provided the aggregation framework query where you can use dot notation to project the embedded value and simple get on the java side to read the value. – Shelleyshellfire

Answer 3

In more recent versions of MongoDB and Java, you can make use of the getEmbedded() method on Document where path is your string path to the embedded value. The following shorthand works exactly like the above. Useful for those who only need this in a few places and don't want to write a whole function.

String path = "model1.MondayModel.gtxdotdot.xdotdot";
Double value = document.getEmbedded(Arrays.stream(path.split("\\.")).toList(), Double.class);

I'm not certain on the efficiency of this method, but it certainly is less code.

Telly