Incorrect data sampling using two collections MongoDB - mongodb

Question

ChatsUsers - this collection is needed to control the filtering of messages, for example, if a user has cleared the history of correspondence - it is only for himself, he cleared it. How this works is simple, the start_message_id - current date is specified which lets you know to return messages message collection with create_date greater than or equal to the new start_message_id.
type ChatsUsers struct {
ID string `json:"id" bson:"id"`
ChatID string `json:"chat_id" bson:"chat_id"`
UserID string `json:"user_id" bson:"user_id"`
StartMessageID int64 `json:"start_message_id,omitempty" bson:"start_message_id"`
EndMessageID int64 `json:"end_message_id,omitempty" bson:"end_message_id"`
}
type Message struct {
ID string `json:"id" bson:"id"`
ChatID string `json:"chat_id" bson:"chat_id"`
FromID string `json:"from_id" bson:"from_id"`
CreateDate int64 `json:"create_date" bson:"create_date"`
Body string `json:"body" bson:"body"`
UpdateAt int64 `json:"update_at" bson:"update_at"`
...
}
type Chat struct {
ID string `json:"id" bson:"id"`
Participants []string `json:"participants" bson:"participants"`
LastMessage *Message `json:"last_message,omitempty" bson:"last_message"`
...
}
Which is what I'm trying to do now in the example below. I need to find all last messages $last_message (see aggregation I wrote) for each chat which creation time $last_message.create_date is greater than or equal to chatsusers.start_message_id.
My problem, my option doesn't work the way I want it to, it doesn't select one message, and $gte doesn't work. I may not have written the query quite correctly, but I'm working on it still
My question: how do I get ONE last message for each chat that has a creation time $last_message.create_date greater than or equal to chatsusers.start_message_id?
My try

Answer 1

You are actually very close. Just remember when using sub-pipeline in $lookup, if you need to access values in the "from" collection, you will need to put it in a variable in let clause. (i.e. $chat_user.start_message_id in your case)
When you access the values in "to" collection, you don't need to put the new name of the $lookup result in the sub-pipeline. (i.e. $last_message.create_date should be simply $create_date)
db.chat.aggregate([
{
$match: {
participants: "63ce54460aeee5e72c778d90"
}
},
{
$lookup: {
from: "chatsusers",
localField: "id",
foreignField: "chat_id",
as: "chat_user"
}
},
{
$unwind: {
path: "$chat_user"
}
},
{
$lookup: {
from: "message",
localField: "id",
foreignField: "chat_id",
let: {
smid: "$chat_user.start_message_id"
},
pipeline: [
{
$match: {
$expr: {
$gte: [
"$create_date",
"$$smid"
]
}
}
},
{
$sort: {
create_date: -1
}
},
{
$limit: 1
}
],
as: "last_message",
}
},
{
$unwind: {
path: "$last_message",
preserveNullAndEmptyArrays: true
}
},
{
$group: {
_id: "$_id",
"create_date": {
$first: "$create_date"
},
"id": {
$first: "$_id"
},
"last_message": {
$max: "$last_message"
},
"owner_id": {
$first: "$owner_id"
},
"participants": {
$first: "$participants"
},
"title": {
$first: "$title"
},
"type": {
$first: "$type"
},
"unread": {
$first: "$unread"
}
}
},
{
$project: {
id: 1,
title: 1,
create_date: 1,
type: 1,
participants: 1,
owner_id: 1,
last_message: "$last_message",
unread: 1
}
}
])
Mongo Playground