Query by data already in the object - node.js

Question

I'm writing a query that gets data from "coll2" based on data that is inside "coll1".
Coll1 has the following data structure:
{
"_id": "asdf",
"name": "John",
"bags": [
{
"type": "typ1",
"size": "siz1"
},
{
"type": "typ2",
"size": "siz2"
}
]
}
Coll2 has the following data structure:
{
_id: "qwer",
coll1Name: "John",
types: ["typ1", "typ3"],
sizes: ["siz1", "siz4"]
}
{
_id: "zxcv",
coll1Name: "John",
types: ["typ2", "typ3"],
sizes: ["siz1", "siz2"]
}
{
_id: "fghj",
coll1Name: "John",
types: ["typ2", "typ3"],
sizes: ["siz1", "siz4"]
}
I want to get all the documents in coll2 that have the same Type+Size combo as in coll1 using the $lookup stage of the aggregation pipeline. I understand that this can be achieved by using the $lookup pipeline and $expr but I cant seem to figure out how to dynamically make a query to pass into the $match stage.
The output I would like to get for the above data would be:
{
_id: "qwer",
coll1Name: "John",
types: ["typ1", "typ3"],
sizes: ["siz1", "siz4"]
}
{
_id: "zxcv",
coll1Name: "John",
types: ["typ2", "typ3"],
sizes: ["siz1", "siz2"]
}

Answer 1

You can use $lookup to get the data from Col2. Then you need to check if there's any element in Col2 ($anyElemenTrue) that matches with Col1. $map and $in can be used here. Then you just need to $unwind and promote Col2 to root level using $replaceRoot
db.Col1.aggregate([
{
$lookup: {
from: "Col2",
localField: "name",
foreignField: "coll1Name",
as: "Col2"
}
},
{
$project: {
Col2: {
$filter: {
input: "$Col2",
as: "c2",
cond: {
$anyElementTrue: {
$map: {
input: "$bags",
as: "b",
in: {
$and: [
{ $in: [ "$$b.type", "$$c2.types" ] },
{ $in: [ "$$b.size", "$$c2.sizes" ] },
]
}
}
}
}
}
}
}
},
{
$unwind: "$Col2"
},
{
$replaceRoot: {
newRoot: "$Col2"
}
}
])

Answer 2

You are correct in your approach to use $lookup with the pipeline field to filter the input documents in the $match pipeline
The $expr expression should typically follow
"$expr": {
"$and": [
{ "$eq": [ "$name", "$$coll1_name" ] },
{ "$setEquals": [ "$bags.type", "$$types" ] },
{ "$setEquals": [ "$bags.size", "$$sizes" ] }
]
}
where the first match expression in the $and conditional { "$eq": [ "$name", "$$coll1_name" ] } checks to see if the name field in coll1 collection matches the coll1Name field in the input documents from coll2.
Of course the fields from coll2 should be defined in a variable in the pipeline with the let field for the $lookup pipeline to access them.
The other match filters are basically checking if the arrays are equal where "$bags.type" from coll1 resolves to an array of types i.e. [ "typ1", "typ3" ] for example.
On getting the output field from $lookup which happens to be an array, you can filter the documents in coll2 on that array field where there can be some empty lists as a resul of the above $lookup pipeline $match filter:
{ "$match": { "coll1Data.0": { "$exists": true } } }
Overall your aggregate pipeline operation would be as follows:
db.getCollection('coll2').aggregate([
{ "$lookup" : {
"from": "coll1",
"let": { "coll1_name": "$coll1Name", "types": "$types", "sizes": "$sizes" },
"pipeline": [
{ "$match": {
"$expr": {
"$and": [
{ "$eq": [ "$name", "$$coll1_name" ] },
{ "$setEquals": [ "$bags.type", "$$types" ] },
{ "$setEquals": [ "$bags.size", "$$sizes" ] }
]
}
} }
],
"as": "coll1Data"
} },
{ "$match": { "coll1Data.0": { "$exists": true } } },
{ "$project": { "coll1Data": 0 } }
])