Populate feild then find

Prerequisites

• I have written a descriptive issue title
• I have searched existing issues to ensure the feature has not already been requested

🚀 Feature Proposal

I have the following tables:

And I want to make a query, and filter inside the subscriptions’ collection only for all the users who have the role “client”.

So the query should return:

[
  { 
     id: '1',
     name: 'Ahmad',
     role: 'client',
     balance: 0
  }
]

How to filter in MongoDB for this?

The only solution, unfortunately, is to use .aggregate() with $lookup, then $match, then $unwind, a repetitive pattern that always happen for this scenario.

Motivation

Here’s a similar problem that was solved using aggregate: https://stackoverflow.com/a/74580085/18387350

it has many drawbacks, and it’s not clean and makes the code long.

Example

Instead of making Subscription.find().populate(), let us do the opposite Subscription.populate().find().

Subscription.populate('client').find({
   'client.branch': 'BRANCH_ID'
})

• this should mean, populate first, then filter.

[Uzlopak]

Well i strongly doubt that this is fast. You will complain in the future that your operation is slow. Than you either want to change to a SQL-Server or you want to keep mongo but need to invest time to restructure the data to improve the performance.

[Uzlopak]

Also short reminder what a population on mongoose is. When populating a field mongoose loads the data from the database collection and stores it in memory and replaces the corresponding fields of the parent with the documents.

So this means with your proposol to do first populate then find, that you have basically load both, the parent collection and the referenced collection, and then filter. This obviously wont work for big collections.

If you want to achive later some kind of pagination, you would be forced to definetly load the whole collections, as sorting would be only possible in nodejs.

It is one of the principles of working with databases to let the database do the work and not the logic.

I strongly recommend to read about organizing data in nosql databases.