Mongodb aggregation group by inner array - mongodb

Question

I have aggregated my data to give this output.
[
{
"_id": {
"source": "source_1",
"medium": "medium_1",
"campaign": "campaign_1"
},
"visitors": [
{
"_id": "60073f564d6c915237dbe158",
"location": {
"city": "Miami",
"postal": "33177"
}
},
{
"_id": "60073f564d6c915237dbe158",
"location": {
"city": "Miami",
"postal": "33163"
}
}
]
},
{
"_id": {
"source": "source_2",
"medium": "medium_2",
"campaign": "campaign_2"
},
"visitors": [
{
"_id": "60073f564d6c915237dbe158",
"location": {
"city": "Miami",
"postal": "33177"
}
},
{
"_id": "60073f564d6c915237dbe158",
"location": {
"city": "Miami",
"postal": "33162"
}
}
]
}
]
I want to group inner visitors array and get this output.
[
{
"_id": {
"source": "source_1",
"medium": "medium_1",
"campaign": "campaign_1"
},
"visitors": [
{
"city": "Miami",
"postal": "33177",
"count": 2
},
{
"city": "Miami",
"postal": "33163",
"count": 5
}
]
},
{
"_id": {
"source": "source_2",
"medium": "medium_2",
"campaign": "campaign_2"
},
"visitors": [
{
"city": "Miami",
"postal": "33177",
"count": 1
},
{
"city": "Miami",
"postal": "33163",
"count": 3
}
]
}
]
aggregate pipeline executed on campaigns collection:
[{$match: {
website_id: 1,
$or: [
{
source:{
$regex:/goo/,
$options: 'i'
}
},
{
medium:{
$regex:/goo/,
$options: 'i'
}
},
{
campaign:{
$regex:/goo/,
$options: 'i'
}
}
]
}}, {$addFields: {
visitor_id: {
$toObjectId: "$visitor_id"
}
}}, {$lookup: {
from: 'visitors',
localField: 'visitor_id',
foreignField: '_id',
as: 'visitors'
}}, {$unwind: {
path: '$visitors'
}}, {$group: {
_id: {
source: '$source',
medium: '$medium',
campaign: '$campaign',
},
visitors:{
$push: '$visitors'
}
}}, {$unwind: {
path: '$visitors'
}}, {$group: {
_id: {
'city': '$visitors.location.city',
'postal': '$visitors.location.postal'
},
'count': {
'$sum': 1
}
}}, {$project: {
'_id': 0,
'city': '$_id.city',
'postal': '$_id.postal',
'count': '$count',
'total': {
'$sum': '$count'
}
}}, {$project: {
'city': '$city',
'postal': '$postal',
'count': '$count',
'total': {
'$sum': '$total'
}
}}]

Answer 1

So the idea is first group the visitors by their postal number along with the campaign details to get the count and then aggregate it by only campaign details to accumulate the visitors.
Try this query:
db.campaigns.aggregate([
{
$match: {
// Put your condtions here.
}
},
{
$project: {
source: 1,
medium: 1,
campaign: 1,
visitor_id: 1
}
},
{
$addFields: {
visitor_id: { $toObjectId: "$visitor_id" }
}
},
{
$lookup: {
from: "visitors",
let: { "visitor_id": "$visitor_id" },
pipeline: [
{
$match: {
$expr: { $eq: ["$_id", "$$visitor_id"] }
}
},
{
$project: {
location: {
city: 1,
postal: 1
}
}
}
],
as: "visitor"
}
},
{ $unwind: "$visitor" },
{
$group: {
_id: {
source: "$source",
medium: "$medium",
campaign: "$campaign",
postal: "$visitor.location.postal"
},
visitors: { $push: "$visitor" },
count: { $sum: 1 }
}
},
{
$group: {
_id: {
source: "$_id.source",
medium: "$_id.medium",
campaign: "$_id.campaign"
},
visitors: {
$push: {
city: { $arrayElemAt: ["$visitors.location.city", 0] },
postal: { $arrayElemAt: ["$visitors.location.postal", 0] },
count: "$count"
}
}
}
}
]);

Answer 2

You need to correct group stage,
$group by source, medium, campaign and postal, get first city and count total sum
$group by source, medium, campaign and construct visitors array with required fields
db.campaigns.aggregate([
{ $match: .. } //skipped
{ $addFields: .. }, //skipped
{ $lookup: .. }, //skipped
{ $unwind: .. }, //skipped
{
$group: {
_id: {
source: "$source",
medium: "$medium",
campaign: "$campaign",
postal: "$visitors.location.postal"
},
city: { $first: "$visitors.location.city" },
count: { $sum: 1 }
}
},
{
$group: {
_id: {
source: "$_id.source",
medium: "$_id.medium",
campaign: "$_id.campaign"
},
visitors: {
$push: {
city: "$city",
postal: "$_id.postal",
count: "$count"
}
}
}
}
])
Playground