MongoDB update latest subdocument

Question

here is my mongo document..

{
    "_id" : ObjectId("5a69d0acb76d1c2e08e4ccd8"),
    "subscriptions" : [ 
        {
            "sub_id" : "5a56fd399dd78e33948c9b8e",
            "invoice_id" : "5a56fd399dd78e33948c9b8d"
        }, 
        {
            "sub_id" : "5a56fd399dd78e33948c9b8e"
        }
    ]
}

i want to update and upsert invoice_id into last element of sub-array.. i have tried..
sort: {$natural: -1},
subscription.$.invoice

what i want it to be is....

{
    "_id" : ObjectId("5a69d0acb76d1c2e08e4ccd8"),
    "subscriptions" : [ 
        {
            "sub_id" : "5a56fd399dd78e33948c9b8e",
            "invoice_id" : "5a56fd399dd78e33948c9b8d"
        }, 
        {
            "sub_id" : "5a56fd399dd78e33948c9b8e",
            "invoice_id" : "5a56fd399dd78e33948c9b8f"
        }
    ]
}

Answer 1

While there are ways to get the last array element, like Saravana shows in her answer, I don't recommend doing it that way because it introduces race conditions. For example, if two subs are added simultaneously, you can't depend on which one is 'last' in the array.

If an invoice_id has to be tied to a specific sub_id, then it's far better to query and find that specific element in the array, then add the invoice_id to it.

In the comments, the OP indicated that the current order of operations is 1) add sub_id, 2) insert the invoice record into the INVOICE collection and get the invoice_id, 3) add the invoice_id into the new subscription.

However, if you already have the sub_id, then it's better to re-order your operations this way: 1) insert the invoice record and get the invoice_id 2) add both sub_id and invoice_id with a single operation.

Doing this improves performance (eliminates the second update operation), but more importantly, eliminates race conditions because you're adding both sub_id and invoice_id at the same time.

Answer 2

we can get the document and update last element by index

> var doc = db.sub.findOne({"_id" : ObjectId("5a69d0acb76d1c2e08e4ccd8")})

> if ( doc.subscriptions.length - 1 >= 0 ) 
    doc.subscriptions[doc.subscriptions.length-1].invoice_id="5a56fd399dd78e33948c9b8f"

> db.sub.update({_id:doc._id},doc)
WriteResult({ "nMatched" : 1, "nUpserted" : 0, "nModified" : 1 })

or write an aggregation pipeline to form the document and use it for update

db.sub.aggregate(
    [
        {$match : { "_id" : ObjectId("5a69d0acb76d1c2e08e4ccd8") }},
        {$addFields : { last : { $subtract : [{$size : "$subscriptions"},1]}}},
        {$unwind : { path :"$subscriptions" , includeArrayIndex : "idx"}},
        {$project : { "subscriptions.sub_id" : 1, 
                    "subscriptions.invoice_id" : {
                        $cond : {
                            if: { $eq: [ "$idx", "$last" ] },
                            then: "5a56fd399dd78e33948c9b8f",
                            else: "$$REMOVE"
                        }
                    }
                }
        },
        {$group : {_id : "$_id", subscriptions : {$push : "$subscriptions"}}}
    ]
).pretty()

result doc

{
    "_id" : ObjectId("5a69d0acb76d1c2e08e4ccd8"),
    "subscriptions" : [
        {
            "sub_id" : "5a56fd399dd78e33948c9b8e"
        },
        {
            "sub_id" : "5a56fd399dd78e33948c9b8e",
            "invoice_id" : "5a56fd399dd78e33948c9b8f"
        }
    ]
}