Mongodb user table with friends - arrays

Question

I have a USER table with documents:
{
_id: 1,
name: 'funny-guy43',
image: '../../../img1.jpg',
friends: [2, 3]
},
{
_id: 2,
name: 'SurfinGirl3',
image: '../../../img2.jpg',
friends: []
},
{
_id: 3,
name: 'FooBarMan',
image: '../../../img3.jpg',
friends: [2]
}
friends is an array of USER _ids. (1) I want to get user by _id, (2) look at his friends and (3) query the USER table with the friend ids to return all friends.
for example, find user 1, query the table based on his friends 2 and 3, and return 2 and 3.
Can I do that in one transaction? Or do I query the table to get user array of friends, then query the table again with array of friends ids.
I'm using .Net Core if that matters.
I am very open to alternative approaches as well.

Answer 1

It is, in fact, possible to do this in one transaction. Or, to be more exact, in one aggregation.
I would first split the users into 2 different subsets, one called searched_user and the other other_users, where searched_user will have only the user we are searching for and other_users will have everyone else. We can do that using $facet. Here is the idea:
{
"$facet": {
"searched_user": [
{
$match: {
_id: 1
}
}
],
"other_users": [
{
$match: {
_id: {
$ne: 1
}
}
}
]
}
}
Once they are separated like this, we can search the other_users subset using the friend ids from the searched_user. So here is the full aggregation:
db.collection.aggregate([
{
"$facet": {
"searched_user": [
{
$match: {
_id: 1
}
}
],
"other_users": [
{
$match: {
_id: {
$ne: 1
}
}
}
]
}
},
{
"$unwind": "$searched_user"
},
{
$project: {
user_friends: {
$filter: {
input: "$other_users",
as: "other_users",
cond: {
$in: [
"$$other_users._id",
"$searched_user.friends"
]
}
}
}
}
}
])
Here we are looking for user 1 and the result will be user 1's friends.
[
{
"user_friends": [
{
"_id": 2,
"friends": [],
"image": "../../../img2.jpg",
"name": "SurfinGirl3"
},
{
"_id": 3,
"friends": [
2
],
"image": "../../../img3.jpg",
"name": "FooBarMan"
}
]
}
]
Playground: https://mongoplayground.net/p/-8pNnQXg8r6

Answer 2

You can achieve this by using lookup in aggregation, Tried it with MongoDB version v4.2.11.
db.users.aggregate([
{
'$match': {
'_id': 1,
}
},
{
'$lookup': {
'from' : 'users',
'let' : {
'friendIds': '$friends',
},
'pipeline': [
{
'$match':{
'$expr': {'$in': [ '$_id', '$$friendIds']}
}
}
],
'as': 'friendsArr'
}
}
])
Result:
[
{
"_id" : 1,
"name" : "funny-guy43",
"image" : "../../../img1.jpg",
"friends" : [
2,
3
],
"friendsArr" : [
{
"_id" : 2,
"name" : "SurfinGirl3",
"image" : "../../../img2.jpg",
"friends" : [ ]
},
{
"_id" : 3,
"name" : "FooBarMan",
"image" : "../../../img3.jpg",
"friends" : [
2
]
}
]
}
]