Double aggregation with distinct count in MongoDB - mongodb

Question

We have a collection which stores log documents.
Is it possible to have multiple aggregations on different attributes?
A document looks like this in it's purest form:
{
_id : int,
agent : string,
username: string,
date : string,
type : int,
subType: int
}
With the following query I can easily count all documents and group them by subtype for a specific type during a specific time period:
db.logs.aggregate([
{
$match: {
$and : [
{"date" : { $gte : new ISODate("2020-11-27T00:00:00.000Z")}}
,{"date" : { $lte : new ISODate("2020-11-27T23:59:59.000Z")}}
,{"type" : 906}
]
}
},
{
$group: {
"_id" : '$subType',
count: { "$sum": 1 }
}
}
])
My output so far is perfect:
{
_id: 4,
count: 5
}
However, what I want to do is to add another counter, which will also add the distinct count as a third attribute.
Let's say I want to append the resultset above with a third attribute as a distinct count of each username, so my resultset would contain the subType as _id, a count for the total amount of documents and a second counter that represents the amount of usernames that has entries. In my case, the number of people that somehow have created documents.
A "pseudo resultset" would look like:
{
_id: 4,
countOfDocumentsOfSubstype4: 5
distinctCountOfUsernamesInDocumentsWithSubtype4: ?
}
Does this makes any sense?
Please help me improve the question as well, since it's difficult to google it when you're not a MongoDB expert.

Answer 1

You can first group at the finest level, then perform a second grouping to achieve what you need:
db.logs.aggregate([
{
$match: {
$and : [
{"date" : { $gte : new ISODate("2020-11-27T00:00:00.000Z")}}
,{"date" : { $lte : new ISODate("2020-11-27T23:59:59.000Z")}}
,{"type" : 906}
]
}
},
{
$group: {
"_id" : {
subType : "$subType",
username : "$username"
},
count: { "$sum": 1 }
}
},
{
$group: {
"_id" : "$_id.subType",
"countOfDocumentsOfSubstype4" : {$sum : "$count"},
"distinctCountOfUsernamesInDocumentsWithSubtype4" : {$sum : 1}
}
}
])
Here is the test cases I used:
And here is the aggregate result: