28

I have document structure like :

{
    map: 'A',
    points: [
        {
            type: 'type1',
            distanceToSpawn: 110
        },
        {
            type: 'type4',
            distanceToSpawn: 40
        },
        {
           type: 'type6',
           distanceToSpawn: 30
        }
    ]
},
{
    map: 'B',
    points: [
        {
            type: 'type1',
            distanceToSpawn: 100
        },
        {
            type: 'type2',
            distanceToSpawn: 60
        },
        {
            type: 'type3',
            distanceToSpawn: 25
        }
    ]
},
{
    map: 'C',
    points: [
        {
            type: 'type2',
            distanceToSpawn: 90
        },
        {
            type: 'type3',
            distanceToSpawn: 1
        },
        {
            type: 'type6',
            distanceToSpawn: 76
        }
    ]
}

I want to get all maps having point type type1 sorted by the distanceToSpawn in ascending order.

Expected result :

{
    map: 'B',
    points: [
        {
            type: 'type1',
            distanceToSpawn: 100
        }
    ]
},
{
    map: 'A',
    points: [
        {
            type: 'type1',
            distanceToSpawn: 110
        }
    ]
}

I've tried something like :

db.maps.find({'points.type': {$eq : 'type1'}}, {map: 1, 'points.$':1}).sort({'points.distanceToSpawn': 1}).limit(10)

But this thing not sorting maps by ascending order.

Thanks.

1 Answer 1

29

You cannot do that with arrays, and the main problem here is because you want the "sort" to happen on the matched element. If you want to sort results like this then you need to use .aggregate() instead. Either as:

For modern MongoDB versions:

db.maps.aggregate([
    { "$match": { "points.type": "type1" }},
    { "$addFields": {
        "order": {
            "$filter": {
              "input": "$points",
              "as": "p",
              "cond": { "$eq": [ "$$p.type", "type1" ] }
            }
        }
    }},
    { "$sort": { "order": 1 } }
])

For MongoDB 2.6 to 3.0

db.maps.aggregate([
   { $match: { 'points.type': 'type1' } },
    {
     $project: {
       points: {
        $setDifference: [
          {
            $map: {
              input: '$points',
              as: 'p',
              in: {
                $cond: [
                  { $eq: ['$$p.type', 'type1'] },
                  '$$p',
                  false,
                ]
              }
            }
          },
          [false]
        ]
      }
    }
  },
  { $sort: { 'points.distanceToSpawn': 1 } },
]);

Or less efficiently in versions prior to MongoDB 2.6:

db.maps.aggregate([
    { "$match": { "points.type": "type1" }},
    { "$unwind": "$points" },
    { "$match": { "points.type": "type1" }},
    { "$group": {
        "_id": "$_id",
        "points": { "$push": "$points" }
    }},
    { "$sort": { "points.ditanceToSpawn": 1 } }         
])

That's the only way to match the correct elements and have them considered in a "sort" operation. The default cursor sort will just otherwise consider the values for the field in the array elements that do not match your selected "type" instead.

Sign up to request clarification or add additional context in comments.

4 Comments

Keep in mind that when using $unwind operator before the $sort operator it cannot take advantages of an index. (I had the same issues trying to sort by array filds using always the first item of the array for comparism.)
For the "modern MongoDB versions" example, should the sort be on order.distanceToSpawn instead of just order? Or if not, why?
What index would you create for this sorting ?
This is a bad solution because due to the lack of an index, all data will be load into RAM and then sorted.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.