CODEWITHSUNDEEP
node.jsmongodbmongoose
hielsnoppe
hielsnoppe

Reputation: 2869

Modelling reference to embedding document using Mongoose

I am modelling two types of events (events and subevents) in a MongoDB like this:

var EventSchema = mongoose.Schema({
    'name' : String,
    'subEvent' : [ SubeventSchema ]
});

var SubeventSchema = mongoose.Schema({
    'name' : String
});

Now when I query a subevent I want to be able to also retrieve data about its corresponding superevent, so that some example data retrieved using Mongoose population feature could look like this:

EventModel.findOne({
    name : 'Festival'
})
.populate('subEvent')
.execute(function (err, evt) { return evt; });

{
    name : 'Festival',
    subEvent: [
        { name : 'First Concert' },
        { name : 'Second Concert' }
    ]
}

EventModel.findOne({
    'subEvent.name' : 'FirstConcert'
}, {
    'subEvent.$' : 1
})
.populate('superEvent') // This will not work, this is the actual problem of my question
.execute(function (err, subevt) { return subevt; });

{
    name: 'First Concert',
    superEvent: {
        name: 'Festival'
    }
}

A solution I can think of is not to embed but to reference like this:

var EventSchema = mongoose.Schema({
    'name' : String,
    'subEvent' : [ {
        'type' : mongoose.Schema.Types.ObjectId,
        'ref' : 'SubeventSchema'
    } ]
});

var SubeventSchema = mongoose.Schema({
    'name' : String,
    'superEvent' : {
        'type' : mongoose.Schema.Types.ObjectId,
        'ref' : 'EventSchema'
    }
});

I am looking for a solution based on the first example using embedded subevents, though. Can this be achieved and in case yes, how?

Upvotes: 1

Views: 995

Answers (2)

wdberkeley
wdberkeley

Reputation: 11671

I think your mental model of document embedding isn't correct. The major misunderstanding (and this is very common) is that you "query a subevent" (query an embedded document). According to your current Event schema, a Subevent is just a document embedded in an Event document. The embedded SubEvent is not a top-level document; it's not a member of any collection in MongoDB. Therefore, you don't query for it. You query for Events (which are the actual collection-level documents in your schema) whose subEvents have certain properties. E.g. one way people translate the query

db.events.find({ "subEvent" : { "name" : "First Concert" } })

into plain English is as "find all the subevents with the name "First Concert". This is wrong. The right translation is "find all events that have at least one subevent whose name is "First Concert" (the "at least one" part depends on knowledge that subEvent is an array).

Coming back to the specific question, you can hopefully see now that trying to do a populate of a "superevent" on a subevent makes no sense. Your queries return events. The optimal schema, be it subevents embedded in events, one- or two-way references between events and subevents documents in separate collections, or events denormalized into the constituent subevent documents, cannot be determined from the information in the question because the use case is not specified.

Upvotes: 2

Neil Lunn
Neil Lunn

Reputation: 151112

Perhaps this is a situation where you need to modify your thinking rather than the schema itself. Mongoose .populate() supports the basic ideas of MongoDB "projection", or more commonly referred to as "field selection". So rather than try to model around this, just select the fields you want to populate.

So your second schema form is perfectly valid, just change how you populate:

EventModel.find({}).populate("subEvent", "name").execute(function(err,docs) {

  // "subevent" array items only contain "name" now

});

This is actually covered in the Mongoose documentation under the "populate" section.

Upvotes: 0

Related Questions