How to use Mongo Aggregation to limit results around a given input? - mongodb

Question

I looked through the pipeline stages docs, but did not see how to do this.
Suppose you have a user, and each user has points.
User Points
A 22
B 11
C 15
D 7
So, we use '$sort': { points: -1 } to order the users by points.
Is it possible to use a Mongo Aggregation Stage to find the users before and after a given user?
So, given user C (by id), it would return [A, C, B].

Answer 1

Very interesting question. Maybe exists any better solution.
Disclaimer: I assume the user points is unique
We can use $facet to get expected result, but at high cost (very large query)
db.collection.aggregate([
{
$facet: {
"givenUser": [
{
$match: {
"user": "C"
}
}
],
"allUser": [
{
$sort: {
"Points": -1
}
}
],
"orderedPoints": [
{
$sort: {
"Points": -1
}
},
{
$group: {
_id: null,
Points: {
$push: "$Points"
}
}
},
{
$unwind: "$Points"
}
]
}
},
{
$project: {
allUser: 1,
currIndex: {
$indexOfArray: [
"$orderedPoints.Points",
{
$arrayElemAt: [
"$givenUser.Points",
0
]
}
]
},
beforeIndex: {
$add: [
{
$indexOfArray: [
"$orderedPoints.Points",
{
$arrayElemAt: [
"$givenUser.Points",
0
]
}
]
},
-1
]
},
afterIndex: {
$add: [
{
$indexOfArray: [
"$orderedPoints.Points",
{
$arrayElemAt: [
"$givenUser.Points",
0
]
}
]
},
1
]
}
}
},
{
$project: {
result: [
{
$arrayElemAt: [
"$allUser",
{
$cond: {
if: {
$lt: [
"$beforeIndex",
0
]
},
then: 999,
else: "$beforeIndex"
}
}
]
},
{
$arrayElemAt: [
"$allUser",
"$currIndex"
]
},
{
$arrayElemAt: [
"$allUser",
"$afterIndex"
]
}
]
}
}
])
[
{
"result": [
{
"Points": 22,
"_id": ObjectId("5a934e000102030405000000"),
"user": "A"
},
{
"Points": 15,
"_id": ObjectId("5a934e000102030405000002"),
"user": "C"
},
{
"Points": 11,
"_id": ObjectId("5a934e000102030405000001"),
"user": "B"
}
]
}
]
MongoPlayground
Steps:
We keep into separate fields:
Given user (C),
Order all users by points
Order all points and store inside array (I wish MongoDB allows find array index by object too)
Now we find given user index, calculate indexes for "before"/"after" players.
Now, we create result with 3 elements (before, current, after).
Note: If given user is first / last, we ensure to return null for before / after items.