使用流星中的mongo聚合来总计解锁/锁定的集合

Question

So, I have one collection that stores docs related to a user with a structure like:

{_id: "hofjew233332j4", userId: "fhewojfw34324", achievementUnlocked: true };

What I want to do is use the aggregate and underscore to be able to group the docs by user id and then calculate what percentage of their records have unlocked set to true such that a resulting doc would look like:

{_id: "fhewojfw34324(userId)", unlockPercentage: 40 (achievementUnlocked: true / all docs) }

Would I be able to do this while only retrieving the documents once?

Answer 1

First group by count achievementUnlocked true count and used after that in project to calculate percentage as using below aggregation:

db.collectionName.aggregate([{
    "$group": {
        "_id": "$userId",
        "achievementUnlockedTrueCount": {
            "$sum": {
                "$cond": {
                    "if": {
                        "$eq": ["$achievementUnlocked", true] //count achievementUnlocked = true count 
                    },
                    "then": 1,
                    "else": 0
                }
            }
        },
        "totalCount": {
            "$sum": 1 // get total count of grouped documents 
        }
    }
}, {
    "$project": {
        "unlockPercentage": {
            "$multiply": [{
                "$divide": ["$achievementUnlockedTrueCount", "$totalCount"] //used this in project to caculate %
            }, 100]
        }
    }
}]).pretty()

Answer 2

I would personally not even bother with aggregation here as the data would seem trivial. It would be far more efficient to maintain an array of "locked" and "unlocked" achievements per user and/or "game" data.

Take a document like this:

{
    "_id": "hofjew233332j4", 
    "userId": "fhewojfw34324",
    "gameId": "v3XWHHvFSHwYxxk6H",
    "achievementsCount": 5,
    "locked": ["One","Two","Four","Five"],
    "lockedCount": 4,
    "unlocked": ["Three"]
    "unlockedCount": 1
}

So you would initialize per user and "game" here with all "locked" achievements normally, but in this case we will show one already within "unlocked". Also note the "count" fields reflect the number of entries present in each array.

To "unlock" another achievement, then you would simply perform an update to remove from the "locked" array and insert into the "unlocked" array, all while maintaining the "count" values:

Achievements.update(
   { 
       "userId": "fhewojfw34324",
       "gameId": "v3XWHHvFSHwYxxk6H",
       "locked": "Four",
       "unlocked": { "$ne": "Four" }
   },
   {
       "$push": { "unlocked": "Four" },
       "$pull": { "locked": "Four" },
       "$inc": {
           "lockedCount": -1,
           "unlockedCount": 1
       }
   }
)

Which alters the document to this state:

{
    "_id": "hofjew233332j4", 
    "userId": "fhewojfw34324",
    "gameId": "v3XWHHvFSHwYxxk6H",
    "achievementsCount": 5,
    "locked": ["One","Two","Five"],
    "lockedCount": 3,
    "unlocked": ["Three","Four"]
    "unlockedCount": 2
}

It's a very simple pattern to follow as each update maintains the correct values and data. If you wanted information such as a "percentage" then this is a simple matter of:

Achievements.aggregate([
    { "$project": {
        "userId": 1,
        "gameId": 1,
        "percentUnlocked": { "$divide": [ "$unlockedCount", "$achivementsCount" ] }
])

Or just apply that math in client code instead.

This model also makes "real" aggregations you might want to do a lot simplier, and with a lot more scope to information. Plus, it's much more efficient to calculate as you go, than to require something to "add up" the data as a separate process.