mongodb aggregate get last object from sub document that mutch a query - mongodb

Question

We have the following set of documents stored in mongodb:
Conversation record:
{_id: "x", lang: "en", timestamp: "", ... }
Each conversation has many processes, each process has set of messages as child-document list.
Process record:
{_id: "y", conversationId: "x", name: "", timestamp: "", messages: [
{
"direction" : "out",
"text" : "How can I help you?",
"timestamp" : ISODate("2019-05-23T11:08:18.423Z"),
"_id" : 3
},
{
"direction" : "out",
"text" : "Hello",
"timestamp" : ISODate("2019-05-23T11:08:17.423Z"),
"_id" : 1
},
{
"direction" : "in",
"text" : "Hi",
"timestamp" : ISODate("2019-05-23T11:08:18.423Z"),
"_id" : 2
}
], completed: "true"}
I need to make aggregate query, and get list of conversations, while each conversation should have list of processes ordered by timestamp, and each process should have only the last message (based on id field) from both "in" and "out" directions.
We need to get something like this:
[
{
conversationId: "x",
timestamp: "",
processes: [
{
_id: "y",
name: "",
timestamp: "",
lastInMessage: {
"direction": "in",
"text": "Hi",
"timestamp": ISODate("2019-05-23T11:08:18.423Z"),
"_id": 2
},
lastOutMessage: {
"direction": "out",
"text": "How can I help you?",
"timestamp": ISODate("2019-05-23T11:08:18.423Z"),
"_id": 3
}
},
{
_id: "y",
name: "",
....
}
]
}
]
The query I tried is:
conversation.aggregate([
{
$match: {
timestamp: query.timestamp
}
},
{
$project: {
_id: "$conversationId",
timestamp: "$timestamp",
conversationId: "$conversationId"
}
},
{
"$lookup":
{
"from": "processes",
"localField": "conversationId",
"foreignField": "conversationId",
"as": "process"
}
},
{
$project: {
_id: "$_id",
timestamp: "$timestamp",
// messages: "$process.messages",
processes: "$process"
}
},
// here I don't know what to do.

Answer 1

You can use $reduce to calculate min and max value of an array and use $map to generate aggregates for all process values:
{
$project: {
_id: "$_id",
timestamp: "$timestamp",
processes: {
$map: {
input: "$processes",
as: "process",
in: {
_id: "$$process._id",
name: "$$process.name",
timestamp: "$$process.timestamp",
lastInMessage: {
$reduce: {
input: "$$process.messages",
initialValue: null,
in: {
$cond: [
{ $and: [ { $eq: [ "$$this.direction", "in" ] }, { $gt: [ "$$this.timestamp", "$$value.timestamp" ] } ] },
"$$this",
"$$value"
]
}
}
},
lastOutMessage: {
$reduce: {
input: "$$process.messages",
initialValue: null,
in: {
$cond: [
{ $and: [ { $eq: [ "$$this.direction", "out" ] }, { $gt: [ "$$this.timestamp", "$$value.timestamp" ] } ] },
"$$this",
"$$value"
]
}
}
}
}
}
}
}
}
MongoDB Playground example