Group By Count on Array MongoDB

Question

This is one out of ~10 million rows in my database:

{
    "_id" : ObjectId("58f569159f809c49ffc5cdbb"),
    "date" : ISODate("2017-03-21T09:10:07.686Z"),
    "p1" : {
        "_id" : 1765906,
        "deck" : [ 
            33, 
            25, 
            59, 
            38, 
            62, 
            3, 
            33, 
            57
        ],
        "crowns" : 3
    },
    "p2" : {
        "_id" : 2520156,
        "deck" : [ 
            25, 
            69, 
            86, 
            8, 
            44, 
            61, 
            69, 
            50
        ],
        "crowns" : 2
    }
}

It is a log of a game battle. The deck is an array which consists of 8 different cards. I am trying to find the card which has the most wins. A winner can be determined by comparing the crowns (I managed to select all documents where player 1 has won).

What I am trying to achieve:

1. Unfortunately I wasn't able to perform a group query which would return what I am looking for (the card with the most wins).

2. I am also trying to find the most successful deck (quantity of wins is enough - the order of the cards specified in that array should be ignored).

What I have tried, but returned an empty error after some time:

db.battle_logs.aggregate([
    {
        $addFields: {
            "player1won": { "$cmp": [ "$p1.crowns", "$p2.crowns" ] }
        }
    },
    { $match: { "player1won": 1 }},
    { 
        $group: 
        {
            _id: "$p1.deck",
            count: { $sum: 1 }
        }
    }
], {
  allowDiskUse:true,
  cursor:{}
 })

I expected to get all deck combinations grouped by its quantity of wins (only player1 wins considered)

Answer 1

As you are using the array p1.deck as _id in your group query, you need to sort the array, otherwise the order of the card will create a new group.

Here is a something that might work for your case:

1)

db.battle_logs.aggregate([
    {
        $addFields: {
            "player1won": { "$cmp": [ "$p1.crowns", "$p2.crowns" ] }
        }
    },
    //Fiter out ties
    { 
        $match: { 
            "$or" : [
                {"player1won": 1},  //Player 1 won
                {"player1won" : -1} //Player 2 won
            ]
        }
    },
    // Get the winning deck
    {
        $project: {
            "deck": { $cond : [ { $gte : [ "$p1.crowns", "$p2.crowns" ]}, "$p1.deck", "$p2.deck" ]} 
        }
    },
    //Unwind the deck to get every card
    { 
        $unwind : "$deck"
    },
    //Get count of each card
    { 
        $group : { 
            "_id" : "$deck" , 
            "count" : {"$sum" : 1}
        }
    }, 
    // Sort on count
    {
        $sort: {"count" : -1}

    },
    //Get the card with highest count
    { 
        $limit: 1
    }
], {
  allowDiskUse:true,
  cursor:{}
 })

2)

db.battle_logs.aggregate([
    {
        $addFields: {
            "player1won": { "$cmp": [ "$p1.crowns", "$p2.crowns" ] }
        }
    },
    //Fiter out ties
    { 
        $match: { 
            "$or" : [
                {"player1won": 1},  //Player 1 won
                {"player1won" : -1} //Player 2 won
            ]
        }
    },
    // Get the winning deck
    {
        $project: {
            "deck": { $cond : [ { $gte : [ "$p1.crowns", "$p2.crowns" ]}, "$p1.deck", "$p2.deck" ]} 
        }
    },
    //Unwind the deck to get every card
    { 
        $unwind : "$deck"
    },
    //Sort the cards
    { 
        $sort : {"deck": 1}
    },
    //Group sorted cards back in deck
    { 
        $group : { 
            "_id" : "$_id" , 
            "deck" : {"$push" : "$deck"}
        }
    }, 
    { 
        $group: 
        {
            _id: "$deck",
            count: { $sum: 1 }
        }
    }
], {
  allowDiskUse:true,
  cursor:{}
 })