MongoDB - remove unique documents

I need to remove documents those are unique with a field value. I am running a query to find out documents by some conditions. Here is the query -

db.contents.aggregate([
{
$match: {
$or: [
{
$and: [
{ widgetId: new ObjectId("648d461dae1b9fb2af57fbb2") },
{ value: 'Yes' }
]
},
{
$and: [
{ widgetId: new ObjectId("648d4111ae1b9fb2af57f8f9") },
{
value: "20"
}
]
}
]
}
}
])

I get following documents from contents collection after running the query -

[{
_id: ObjectId("648d50d3221f77c425aac162"),
label: 'Featured Product',
itemId: ObjectId("648d50d3221f77c425aac146"),
slug: 'featured-product',
value: 'Yes',
widgetId: ObjectId("648d461dae1b9fb2af57fbb2")
}
{
_id: ObjectId("648d5131221f77c425aac1b9"),
label: 'Featured Product',
itemId: ObjectId("648d5131221f77c425aac1b0"),
slug: 'featured-product',
value: 'Yes',
widgetId: ObjectId("648d461dae1b9fb2af57fbb2")
}
{
_id: ObjectId("648d5199221f77c425aac223"),
label: 'Featured Product',
itemId: ObjectId("648d5199221f77c425aac21a"),
slug: 'featured-product',
value: 'Yes',
widgetId: ObjectId("648d461dae1b9fb2af57fbb2")
}
{
_id: ObjectId("648d51d1221f77c425aac28b"),
label: 'Discount',
itemId: ObjectId("648d51d1221f77c425aac284"),
slug: 'discount',
value: '20',
widgetId: ObjectId("648d4111ae1b9fb2af57f8f9")
}
{
_id: ObjectId("648d51d1221f77c425aac28d"),
label: 'Featured Product',
itemId: ObjectId("648d51d1221f77c425aac284"),
slug: 'featured-product',
value: 'Yes',
widgetId: ObjectId("648d461dae1b9fb2af57fbb2")
}]

Now, in the same query I need to remove all documents those have unique itemId. So, the result output should be -

[{
_id: ObjectId("648d51d1221f77c425aac28b"),
label: 'Discount',
itemId: ObjectId("648d51d1221f77c425aac284"),
slug: 'discount',
value: '20',
widgetId: ObjectId("648d4111ae1b9fb2af57f8f9")
}
{
_id: ObjectId("648d51d1221f77c425aac28d"),
label: 'Featured Product',
itemId: ObjectId("648d51d1221f77c425aac284"),
slug: 'featured-product',
value: 'Yes',
widgetId: ObjectId("648d461dae1b9fb2af57fbb2")
}]

I think, I need to apply filter to perform the task but can't find out a solution. There might be other approaches to that.

I think we might be able to accomplish that by using below query:

db.collection.aggregate([
// grouping based on itemId ;
{
$group: {
_id: "$itemId",
count: {
$count: {}
}
}
},
//  remove those who have count < 2 (unique elements)
{
$match: {
"count": {
$gte: 2
},
}
},
// populate documents
{
$lookup: {
from: "collection",
localField: "_id",
foreignField: "itemId",
as: "fromItems"
}
},
// level up 'fromItems'
{
$unwind: "$fromItems"
},
{
$replaceRoot: {
newRoot: "$fromItems"
}
}
])

The above code runs in 3 phases,

First we perform grouping of documents, in order to observe which one of them are unique. We then remove those documents.

In the second phase, we have a collection of non-unique documents but with no data but only their ids, we need to populate them. So, we then perform join on them with itself (using $lookup).

Finally, we just move fromItems a level up, effectively making it a document instead of an embedded document.

You can test this out to see if its working or not.