Multiple mongoose count() queries to a MongoDB

Question

first: me -> MongoNoob and I know this has already been asked in one or the other way, but I haven't found anything specific until now. Let's say I have two Moongoose Models described like this:

var pollSchema = mongoose.Schema({
    title: String,
    choices: [{
        content: String
    }]
});
var choiceSchema = mongoose.Schema({
    poll_id: mongoose.Schema.ObjectId,
    option: Number
});

A UI shows the poll and when a user selects a choice, it is written into the choiceSchema model. Now I would like to create a 'statistic', telling me how many users selected option 1, option 2, option 3,.... I could simply fetch all choices for a poll with find and generate the statistic in server code, but if I had a million user choices, I would have to deal with an array of the same size. This cannot be right. I could however generate a query and use the count() method:

var query = Choice.find({poll_id: someId}),
query.where('option', 1);
var resultForOption1;
query.count(function(err, count)) {
    resultForOption1 = count;
});

How would I do this for multiple options and 'join' (haha) the results into an array? Since this is all asynchronous I would have nest the calls, but that is not an option for a variable number of queries.

Do I miss the wood for the trees :-)? Can somebody point me in the right direction?

BR, Daniel

Answer 1

You can use aggregate to do this:

Choice.aggregate(
    {$group: {_id: {poll_id: '$poll_id', option: '$option'}, count: {$sum: 1}}}
).exec(...);

This will group the choices collection docs by poll_id and option and count the occurrences of each, giving output that looks like:

{
  "result": [
    {
      "_id": {
        "poll_id": 2,
        "option": 3
      },
      "count": 1
    },
    {
      "_id": {
        "poll_id": 1,
        "option": 2
      },
      "count": 2
    },
    {
      "_id": {
        "poll_id": 2,
        "option": 2
      },
      "count": 1
    },
    {
      "_id": {
        "poll_id": 1,
        "option": 1
      },
      "count": 1
    }
  ],
  "ok": 1
}

You can then use subsequent $group and $project stages in your aggregate pipeline to further group and reshape the generated docs as needed.

Answer 2

If you want just iterate over the count queries and do it easy and clean you can use a lib like co or even better, async:

async.map({ "option":1 },{ "option":2 }], Choice.find, function(e, r){
    // And you continue from here
});

Another option -better in my opinion-. is to use aggregate queries . You will learn more about mongoDB and should be more efficient. In the link I send to you the first example is almost what you want to do:

_{(source: mongodb.org )}