JIYIK CN >

Today, we will learn how to project nested fields when querying data in MongoDB using the $projectand $unsetaggregation stages, forEach()loops, and methods.mapReduce()

MongoDB item nested fields

In MongoDB, we can find()retrieve all documents using the retrieve method, but what if we want to access only specific nested fields. This is where we use projections.

We can project nested fields in various ways. Here, we will understand the following solutions to project nested fields.

To learn the above method, let us create a nestedcollection called which contains one document. You can also use the query given below to get in touch with us.

Sample code:

// MongoDB version 5.0.8

> db.nested.insertOne(
    {
        "name": {
            "first_name": "Mehvish",
            "last_name": "Ashiq",
         },
         "contact": {
            "phone":{"type": "manager", "number": "123456"},
            "email":{ "type": "office", "mail": "delfstack@example.com"}
         },
         "country_name" : "Australien",
         "posting_locations" : [
             {
                 "city_id" : 19398,
                 "city_name" : "Bondi Beach (Sydney)"
             },
             {
                  "city_id" : 31101,
                  "city_name" : "Rushcutters Bay (Sydney)"
             },
             {
                  "city_id" : 31022,
                  "city_name" : "Wolly Creek (Sydney)"
             }
          ],
          "regions" : {
              "region_id" : 796,
              "region_name" : "Australien: New South Wales (Sydney)"
          }
    }
);

db.nested.find().pretty();View the inserted data on the mongo shell using .

$projectProjecting nested fields in MongoDB using the Aggregate stage

Sample code:

// MongoDB version 5.0.8

> var current_location = "posting_locations";
> var project = {};
> project["id"] = "$"+current_location+".city_id";
> project["name"] = "$"+current_location+".city_name";
> project["regions"] = 1;

> var find = {};
> find[current_location] = {"$exists":true};

> db.nested.aggregate([
    { $match : find },
    { $project : project }
]).pretty()

Output:

{
        "_id" : ObjectId("62a96d397c7e3688aea26d0d"),
        "regions" : {
                "region_id" : 796,
                "region_name" : "Australien: New South Wales (Sydney)"
        },
        "id" : [
                19398,
                31101,
                31022
        ],
        "name" : [
                "Bondi Beach (Sydney)",
                "Rushcutters Bay (Sydney)",
                "Wolly Creek (Sydney)"
        ]
}

Here, we posting_locationssave the first level field named current_locationin a variable named .

We then use this variable to access city_idand city_nameand store them projectin the object, using bracket notation to projectcreate properties for the object. Additionally, we regionsstore the field in project["regions"].

Next, we have another findobject called , aggregate()which we will use in the method to match documents. In aggregate()the method, we use $matchthe stage to match documents and $projectto project fields, whether nested or at the first level.

We use $projectto specify the fields to be displayed in the output. If we only want to project the specified nested fields without any filtering query, we can use the following solution.

Sample code:

// MongoDB version 5.0.8

> var current_location = "posting_locations";
> db.nested.aggregate({
    $project: {
         "_id": 0,
         "city_id": "$" + current_location + ".city_id",
         "city_name": "$" + current_location + ".city_name",
         "regions": 1
    }
}).pretty();

Output:

{
        "regions" : {
                "region_id" : 796,
                "region_name" : "Australien: New South Wales (Sydney)"
        },
        "city_id" : [
                19398,
                31101,
                31022
        ],
        "city_name" : [
                "Bondi Beach (Sydney)",
                "Rushcutters Bay (Sydney)",
                "Wolly Creek (Sydney)"
        ]
}

Use $unsetthe aggregation stage to get nested fields in MongoDB excluding specified fields

Sample code:

// MongoDB version 5.0.8

> db.nested.aggregate({
        $unset: ["posting_locations.city_id", "contact", "regions", "name", "_id"]
}).pretty()

Output:

{
        "country_name" : "Australien",
        "posting_locations" : [
                {
                        "city_name" : "Bondi Beach (Sydney)"
                },
                {
                        "city_name": "Rushcutters Bay (Sydney)"
                },
                {
                        "city_name": "Wolly Creek (Sydney)"
                }
        ]
}

Here, we use $unsetthe operator, which is used to remove a specified field or array of fields.

Remember that we use dot notation to specify an embedded document or array of documents. If the given field does not exist, $unsetthe operator does nothing.

When we $match elements of an array using , $unsetthe operator replaces the matched elements with nullinstead of removing them from the array. This behavior helps in keeping element positions and array sizes consistent.

forEach()Get nested fields in MongoDB using loop

Sample code:

// MongoDB version 5.0.8

> var bulk = db.newcollection.initializeUnorderedBulkOp(),
   counter = 0;

> db.nested.find().forEach(function(doc) {
    var document = {};
    document["name"] = doc.name.first_name + " " + doc.name.last_name;
    document["phone"] = doc.contact.phone.number;
    document["mail"] = doc.contact.email.mail;
    bulk.insert(document);
    counter++;
    if (counter % 1000 == 0) {
        bulk.execute();
        bulk = db.newcollection.initializeUnorderedBulkOp();
    }
});

> if (counter % 1000 != 0) { bulk.execute(); }

You will see something similar to the following.

BulkWriteResult({
        "writeErrors" : [ ],
        "writeConcernErrors" : [ ],
        "nInserted" : 1,
        "nUpserted" : 0,
        "nMatched" : 0,
        "nModified" : 0,
        "nRemoved" : 0,
        "upserted" : [ ]
})

Next, execute the following command on your mongo shell to view the projected fields.

// MongoDB version 5.0.8

> db.newcollection.find().pretty();

Output:

{
        "_id" : ObjectId("62a96f2d7c7e3688aea26d0e"),
        "name" : "Mehvish Ashiq",
        "phone" : "123456",
        "mail" : "delfstack@example.com"
}

To learn from this sample code, suppose we want to get certain nested fields and insert them into a new collection. Here, inserting the transformed fields as documents into the new collection may nestedaffect our operation depending on the size of the collection.

bulk insertWe can avoid this slow insert performance by using the new out-of-order API. It will simplify insert operations by sending them in batches and give us real-time feedback on whether the operation succeeded or failed.

Therefore, we use bulk insertthe API to insert the required data structure newcollectioninto the collection, where brand new documents will be created using a loop nested of the collection cursor forEach(). To create new attributes, we use the bracket notation.

For this code, we assume that there is a large amount of data. Therefore, we will 1000send the operations in batches of to the server to perform bulk insert operations.

As a result, it provides us with good performance because instead of sending every request to the server, we send it once every 1000 requests.

mapReduce()Projecting nested fields in MongoDB using the

Sample code:

// MongoDB version 5.0.8

> function map() {
    for(var i in this.posting_locations) {
         emit({
             "country_id" : this.country_id,
             "city_id" : this.posting_locations[i].city_id,
             "region_id" : this.regions.region_id
         },1);
    }
}

> function reduce(id,docs) {
      return Array.sum(docs);
}

> db.nested.mapReduce(map,reduce,{ out : "map_reduce_output" } )

Now, run the following query to see the output.

// MongoDB version 5.0.8
> db.map_reduce_output.find().pretty();

Output:

{
        "_id" : {
                "country_id" : undefined,
                "city_id" : 19398,
                "region_id" : 796
        },
        "value" : 1
}
{
        "_id" : {
                "country_id" : undefined,
                "city_id" : 31022,
                "region_id" : 796
        },
        "value" : 1
}
{
        "_id" : {
                "country_id" : undefined,
                "city_id" : 31101,
                "region_id" : 796
        },
        "value" : 1
}

For this sample code, we use the map-reduce mapReduce()function to perform a map-reduce on nestedall the documents of the collection. To do this, we must follow a three-step process briefly explained below.