Mongo aggregation in java: group with multiple fields

Question

I'm trying to perform an aggregation operation using in Java using the mongo-java-driver. I've performed some other find operations, but I'm unable to do the following aggregation correctly in Java:

db.I1.aggregate([
    { "$match": { "ci": 862222} },
    { "$match": { "gi": { "$ne": null } }},
    { "$group": {
        "_id": {
            "ci": "$ci",
            "gi": "$gi",
            "gn": "$gn",
            "si": "$si"
        }
    }},
    { "$group": {
        "_id": {
            "ci": "$_id.ci",
            "gi": "$_id.gi",
            "gn": "$_id.gn"
        },
        "sn": { "$sum": 1 }
    }},
    { "$sort" : { "_id.gi" : 1}}
])

I've tried several ways and methods to perform that aggregation in Java, but I'm unable to include the group fields "ci", "gi", "gn","si" correctly in the coll.aggregate(asList()) method. What I got so far, is the following:

MongoCollection<Document> coll = mongo.getCollection("I1");

Document matchCourse = new Document("$match",
    new Document("ci", Integer.parseInt(courseid)));

Document matchGroupNotNull = new Document("$match",
    new Document("gi", new Document("$ne", null)));

List<Object> list1 = new BasicDBList();
    list1.add(new BasicDBObject("ci", "$ci"));
    list1.add(new BasicDBObject("gi", "$gi"));
    list1.add(new BasicDBObject("gn", "$gn"));
    list1.add(new BasicDBObject("si", "$si"));

Document group1 = new Document(
    "_id", list1).append("count", new Document("$sum", 1));

List<Object> list2 = new BasicDBList();
list2.add(new BasicDBObject("ci", "$_id.ci"));
list2.add(new BasicDBObject("gi", "$_id.gi"));
list2.add(new BasicDBObject("gn", "$_id.gn"));

Document group2 = new Document(
    "_id", list2).append("sn", new Document("$sum", 1));

Document sort = new Document("$sort", 
    new Document("_id.gi", 1));

AggregateIterable<Document> iterable = coll.aggregate(asList(matchCourse,
    matchGroupNotNull, group1, group2, sort));

I know it's not correct, but I included it to give you an idea of what I am doing. I've googled this in many different ways and read several pages, but I didn't find any solution. The available documentation for MongoDB-Java(1, 2) is too short for me and doesn't include this case.

How can I perform that query in Java? Any help would be appreciated.

Thank you very much!!

Hardunn

Answer 1

Finally I've come to a solution. There were some errors in the question that I posted, as it was the last attempt after reaching some point of desperation, but finally, here is the final solution:

MongoDatabase mongo = // initialize your connection;
Document matches = new Document("$match",
    new Document("gi", new Document("$ne", null))
    .append("ci", Integer.parseInt(courseid)));

Document firstGroup = new Document("$group",
    new Document("_id",
        new Document("ci", "$ci")
        .append("gi", "$gi")
        .append("gn", "$gn")
        .append("si", "$si"))
    .append("count", new Document("$sum", 1)));

Document secondGroup = new Document("$group",
    new Document("_id",
        new Document("ci", "$_id.ci")
        .append("gi", "$_id.gi")
        .append("gn", "$_id.gn"))
    .append("ns", new Document("$sum", 1))); 

Document sort = new Document("$sort", 
    new Document("_id.gi", 1));             

List<Document> pipeline = Arrays.asList(matches, firstGroup,
    secondGroup, sort);
AggregateIterable<Document> cursor = mongo.getCollection("I1")
    .aggregate(pipeline);

for(Document doc : cursor) { // do stuff with doc }

Instead of trying to create lists of key-values, I just appended the elements to the documents. Hope it will be useful for somebody!!

Hardunn

Answer 2

This question is quite old but was the top match on google when I searched so if anyone is looking for a solution to this I managed to do it in the following way

Aggregation.group(Fields.fields()
            .and("field1")
            .and("field2"))
            .first("name")
            .`as`("name")
            .count().`as`("count")

This will produce the following MDB query:

   { "$group" : 
        { "_id" : 
        { "field1" : "$field1", "field2" : "$field2"}, 
        "name" : { "$first" : "$name"}, 
        "count" : { "$sum" : 1}
   }

Dillion

This API looks like it's specific to Spring Data – Following