Update objects with findOneAndUpdate by similar object value

Question

I am trying to update documents values, if they already exist in a collection, if they have the same value for a specific object key.

For example, I have the following documents -

{ eventType: "A", browsers: [ { name: "Chrome", count: 10 } ]}
{ eventType: "B", browsers: [ { name: "Chrome", count: 2 } ]}
{ eventType: "A", browsers: [ { name: "Chrome", count: 5 }, { name: "Safari", count: 8 } ]}

In the end, I want the collection to merge documents by their event type, and merge browsers aggregations by the browser names value (in the example below, the "Chrome" count is combined to 15); if value does not exist, add the new object to the browsers array (In the example below, adding the "Safari" object to the existing browsers aggregation):

{ eventType: "A", browsers: [ { name: "Chrome", count: 15 }, { name: "Safari", count: 8 } ]}
{ eventType: "B", browsers: [ { name: "Chrome", count: 2 } ]}

I want to use the following query for updating and inserting a new document if needed, but I am not sure how the browser aggregation can be made.

const doc = { eventType: "A", browsers: [ { name: "Chrome", count: 5 }, { name: "Safari", count: 8 } ]}

collection.findOneAndUpdate({
    eventType: doc.eventType
  }, {

    // What to do in here?

  }, {
    upsert: true,
    new: true
  })

Is there any option to use a custom function, such as the $accumulator in the collection.aggregate use case?

P.S - I am using mongoose for this code

user_1365220 · Answer

It is not possible to do this in one statement, since your goal is to collect the documents into a new set of documents by eventType, but you can get much of the work done with an aggregation with a merge:

[{$unwind: {
 path: '$browsers',
 includeArrayIndex: 'b',
 preserveNullAndEmptyArrays: true
}}, {$group: {
 _id: '$eventType',
 eventType: {
  $first: '$eventType'
 },
 browsers: {
  $addToSet: '$browsers'
 }
}}, {$addFields: {
 grouped: 1
}}, {$merge: {
 into: 'so_example',
 on: '_id',
 whenMatched: 'replace',
 whenNotMatched: 'discard'
}}]

This would require a second process to a) remove documents that weren't merged and something to reduce the browsers array so that each browser had the sum of the counts.

user_7999670 · Answer

Apparently, it is not possible to do such thing in a single atomic query. I 'solved' it by changing the schema, so the keys will be dynamic, e.g.

{
  browsers: {
    Chrome: {
      count: 15
    }
  }
}

With this implementation, I build the update section of the findOneAndUpdate dynamically, according to the new document that has to be inserted.

Update objects with findOneAndUpdate by similar object value

2 Answers