Question

I’m using GridFS to store images. Each image has a group_id in it’s meta data that basically just defines if they are part of the same original image. So each image, has multiple sizes, the group_id is what groups the same images together.

I want to query this, but grouped by the group_id. So it would return an array of the group_id’s, each with the object’s info (width, id, etc.).

If I just do a find on the records I want, this is what it looks like:

[ { _id: 4f871d4adaf6fa492f000001,
    width: '500',
    group: '1' },
  { _id: 4f871d4adaf6fa492f000004,
    width: '150',
    group: '1' },
  { _id: 4f871d4bdaf6fa492f000007,
    width: '100',
    group: '1' },
  { _id: 4f871d4bdaf6fa492f00000a,
    width: '50',
    group: '1' },
  { _id: 4f871d51daf6faf42e000001,
    width: '500',
    group: '2' },
  { _id: 4f871d52daf6faf42e000004,
    width: '150',
    group: '2' },
  { _id: 4f871d53daf6faf42e000007,
    width: '100',
    group: '2' },
  { _id: 4f871d53daf6faf42e00000a,
    width: '50',
    group: '2' } ]

I’ve looked at group and mapreduce, and have tried out various options but nothing seems to work. Any suggestions or examples would be great!

Thank you!

Answer 1

One possible way to do this is with the $group function in the new aggregation framework (Available in version 2.1.0):
http://www.mongodb.org/display/DOCS/Aggregation+Framework+-+%24group

> db.images.aggregate({$group:{"_id":"$group", "info":{$push:{"width":"$width", "id":"$_id"}}}})
{
    "result" : [
        {
            "_id" : "2",
            "info" : [
                {
                    "width" : "500",
                    "id" : ObjectId("4f871d51daf6faf42e000001")
                },
                {
                    "width" : "150",
                    "id" : ObjectId("4f871d52daf6faf42e000004")
                },
                {
                    "width" : "100",
                    "id" : ObjectId("4f871d53daf6faf42e000007")
                },
                {
                    "width" : "50",
                    "id" : ObjectId("4f871d53daf6faf42e00000a")
                }
            ]
        },
        {
            "_id" : "1",
            "info" : [
                {
                    "width" : "500",
                    "id" : ObjectId("4f871d4adaf6fa492f000001")
                },
                {
                    "width" : "150",
                    "id" : ObjectId("4f871d4adaf6fa492f000004")
                },
                {
                    "width" : "100",
                    "id" : ObjectId("4f871d4bdaf6fa492f000007")
                },
                {
                    "width" : "50",
                    "id" : ObjectId("4f871d4bdaf6fa492f00000a")
                }
            ]
        }
    ],
    "ok" : 1
}
> 

Is this the result that you were hoping to achieve?

A similar result may also be achieved with a Map Reduce Operation:

var map = function(){
    var info = [{"width":this.width, "id":this._id}]; 
    emit(this.group, {"info":info});
}

var reduce = function(key, values){
    var info = [];
    print(key);
    print(values.length);
    for(var v in values){
        print(values[v]);
        for(var i in values[v].info){
            if(info.indexOf(values[v].info[i]) == -1){
                info.push(values[v].info[i]);
            };
        };
    };
    return {"info":info};
}

> db.images.mapReduce(map, reduce, {out:{inline:1}})
{
    "results" : [
        {
            "_id" : "1",
            "value" : {
                "info" : [
                    {
                        "width" : "500",
                        "id" : ObjectId("4f871d4adaf6fa492f000001")
                    },
                    {
                        "width" : "150",
                        "id" : ObjectId("4f871d4adaf6fa492f000004")
                    },
                    {
                        "width" : "100",
                        "id" : ObjectId("4f871d4bdaf6fa492f000007")
                    },
                    {
                        "width" : "50",
                        "id" : ObjectId("4f871d4bdaf6fa492f00000a")
                    }
                ]
            }
        },
        {
            "_id" : "2",
            "value" : {
                "info" : [
                    {
                        "width" : "500",
                        "id" : ObjectId("4f871d51daf6faf42e000001")
                    },
                    {
                        "width" : "150",
                        "id" : ObjectId("4f871d52daf6faf42e000004")
                    },
                    {
                        "width" : "100",
                        "id" : ObjectId("4f871d53daf6faf42e000007")
                    },
                    {
                        "width" : "50",
                        "id" : ObjectId("4f871d53daf6faf42e00000a")
                    }
                ]
            }
        }
    ],
    "timeMillis" : 1,
    "counts" : {
        "input" : 8,
        "emit" : 8,
        "reduce" : 2,
        "output" : 2
    },
    "ok" : 1,
}
> 

Hopefully this will help you to achieve your desired results!