Finding intersection between two object arrays based on field - mongodb

Question

In mongo collection I have documents of following structure.
{
"_id" : "Suzuki",
"qty" : 10,
"plates" : [
{
"rego" : "1QX-WA-123",
"date" : 1516374000000.0
},
{
"rego" : "1QX-WA-456",
"date" : 1513369800000.0
}
],
"accounts" : [
{
"_id" : "23kpi9MD4KnTvnaW7",
"createdAt" : 1513810712802.0,
"date" : 1503446400000.0,
"type" : "Suzuki",
"rego" : "1QX-WA-123",
},
{
"_id" : "2Wqrd4yofvLmqLm5H",
"createdAt" : 1513810712802.0,
"date" : 1501632000000.0,
"type" : "Suzuki",
"rego" : "1QX-WA-111",
}
]
}
I am trying to filter objects in accounts array so that it contains only those objects whose rego exists in plates array.
I tried following query, however, it throws an error: all operands of $setIntersection must be arrays. One argument if of type object.
db.getCollection('dummy').aggregate([{
$project: {
plates: 1,
accounts: 1,
intersect: {
$setIntersection: [
{ $arrayElemAt: [ "$plates", 0 ] },
{ $arrayElemAt: [ "$accounts", 4 ] }
]
}
}
}])
The expected output I am looking for is:
{
"_id" : "Suzuki",
"qty" : 10,
"plates" : [
{
"rego" : "1QX-WA-123",
"date" : 1516374000000.0
},
{
"rego" : "1QX-WA-456",
"date" : 1513369800000.0
}
],
"accounts" : [
{
"_id" : "23kpi9MD4KnTvnaW7",
"createdAt" : 1513810712802.0,
"date" : 1503446400000.0,
"type" : "Suzuki",
"rego" : "1QX-WA-123",
}
]
}

Answer 1

So there are a couple of ways, but what you really are after is simply to $filter instead.
Using $in would likely be the first choice:
db.getCollection('dummy').aggregate([
{ "$addFields": {
"accounts": {
"$filter": {
"input": "$accounts",
"cond": {
"$in": [ "$$this.rego", "$plates.rego" ]
}
}
}
}}
])
Or if you don't have MongoDB 3.4 at least, then using $anyElementTrue:
db.getCollection('dummy').aggregate([
{ "$project": {
"qty": 1,
"plates": 1,
"accounts": {
"$filter": {
"input": "$accounts",
"as": "acc",
"cond": {
"$anyElementTrue": {
"$map": {
"input": "$plates.rego",
"as": "rego",
"in": { "$eq": [ "$$rego", "$$acc.rego" ] }
}
}
}
}
}
}}
])
Or even $setIsSubset:
db.getCollection('dummy').aggregate([
{ "$project": {
"qty": 1,
"plates": 1,
"accounts": {
"$filter": {
"input": "$accounts",
"as": "acc",
"cond": {
"$setIsSubset": [ ["$$acc.rego"], "$plates.rego" ]
}
}
}
}}
])
It's really not a $setIntersection for this type of operation, since that would need a comparison on "just the field values" as a "set", and the output is really just "that" and not the "objects".
You could do something silly with matching array indexes to the produced "set" positions:
db.getCollection('dummy').aggregate([
{ "$addFields": {
"accounts": {
"$map": {
"input": { "$setIntersection": ["$plates.rego", "$accounts.rego"] },
"in": {
"$arrayElemAt": [
"$accounts",
{ "$indexOfArray": [ "$accounts.rego", "$$this" ] }
]
}
}
}
}}
])
But in reality you probably really just want the $filter result as being far more practical. And if you want that output as a "set" then you can simply wrap the $filter output with a $setDifference or like operator to make the entries "unique".
In all variations these return:
{
"_id" : "Suzuki",
"qty" : 10.0,
"plates" : [
{
"rego" : "1QX-WA-123",
"date" : 1516374000000.0
},
{
"rego" : "1QX-WA-456",
"date" : 1513369800000.0
}
],
"accounts" : [
{
"_id" : "23kpi9MD4KnTvnaW7",
"createdAt" : 1513810712802.0,
"date" : 1503446400000.0,
"type" : "Suzuki",
"rego" : "1QX-WA-123"
}
]
}
Showing the items in the "accounts" array "filtered" as matching the respective "rego" amounts from the "plates" array.