$match in mongoose won't query aggregated fields - javascript

I'm trying to query from a field from another collection:
const Order = require("../../models/orders");
const User = require("../../models/user");
module.exports = {
async list(req, res) {
const { page = 1, limit = 10 } = req.query;
let filters = req.query;
filters = {
...filters,
page: undefined,
limit: undefined
};
const myAggregate = Order.aggregate([
{
$lookup: {
from: "users",
localField: "user",
foreignField: "_id",
as: "user"
}
},
{
$unwind: "$user"
},
{
$match: { ...filters }
},
{
$project: {
coupon: 1,
status: 1,
"user._id": 1,
"user.firstName": 1,
"user.lastName": 1,
currency: 1,
rate: 1,
amount: 1,
subtotal: 1,
total: 1,
recipient: 1,
createdAt: 1
}
}
]);
const orders = await Order.aggregatePaginate(myAggregate, { page, limit });
return res.send(orders);
}
};
I'm using a plugin called mongoose-aggregate-paginate-v2, but it just add pagination to the mongoose array, so the problem is with the $match.
The collection Orders stores this object:
{
"_id" : ObjectId("5d447d49a910461fd3b5e155"),
"coupon" : [],
"status" : "Processing",
"user" : ObjectId("5d35f67f1cf34e0018e36d69"),
"currency" : "BRL",
"rate" : 10640,
"amount" : 30,
"subtotal" : 30,
"total" : 30,
"recipient" : ObjectId("5d404b39f005f70419d89ae3"),
"createdAt" : ISODate("2019-07-30T13:59:48.532Z")
}
Which, after the aggregation, turns this:
{
"docs": [
{
"_id": "5d447d49a910461fd3b5e155",
"coupon": [],
"status": "Processing",
"user": {
"_id": "5d35f67f1cf34e0018e36d69",
"firstName": "Boo",
"lastName": "Foo"
},
"currency": "BRL",
"rate": 10640,
"amount": 30,
"subtotal": 30,
"total": 30,
"recipient": "5d404b39f005f70419d89ae3",
"createdAt": "2019-07-30T13:59:48.532Z"
}
],
"totalDocs": 1,
"limit": 10,
"page": 1,
"totalPages": 1,
"pagingCounter": 1,
"hasPrevPage": false,
"hasNextPage": false,
"prevPage": null,
"nextPage": null
}
Inside the docs is the array returned by mongoose. If I make a query of any of the Orders fields, the query returns successful, only with data that corresponds to the query. However, if I try to query the firstName, no matter how I try to format the $match, the query is an empty array.
How should I query aggregated fields?

Related

How to check if an ObjectId is present in array in mongoose?

I am creating a project using Node.js, where I have two models: User and Project.
Here's the Schema for Project model:
const ProjectSchema = new mongoose.Schema({
name: {
type: String,
maxlength: 50,
required: true,
},
description: {
type: String,
maxlength: 800,
required: true
},
contributors: [{
type: mongoose.Schema.Types.ObjectId,
ref: "User"
}],
});
Now, for a project, I want to check if the current user is already present in the project's contributors. If the user is not present, then add the user, otherwise, Ignore.
How can I check that in an efficient manner?
mongoose offers you the aggregate method which does pretty everything, in your case:
const projects = ProjectModel.aggregate([
{ $match: {/* query for matching a small size of projects first(you should definitely do this) */} },
{ $unwind: "$contributors" },
{ $project: { contributor: "$contributors" } }, // This stage is actually optional
{ $match: { "contributor._id": /* contributor id you want to check */ } },
{ $count: "projects" }
]);
Now you can just check the variable and see how many projects have the user in their contributors array
I tested this dataset:
[{
"_id": 1,
"name": "Project1",
"contributors": [{
"_id": 11,
"name": "user1"
}, {
"_id": 22,
"name": "user2"
}]
}, {
"_id": 2,
"name": "Project2",
"contributors": [{
"_id": 22,
"name": "user3"
}, {
"_id": 44,
"name": "user4"
}]
}]
With this query:
const projects = ProjectModel.aggregate([
{ $match: { _id: 1 } },
{ $unwind: "$contributors" },
{ $project: { contributor: "$contributors" } },
{ $match: { "contributor._id": 22 } },
{ $count: "projects" }
]);
The result:
[{
"projects": 1
}]
Playground Link
Query
find the project
check if user _id is on the contributors array
if it is do nothing (keep old value)
else add the new user
replace 2 with the user id and {"_id": 2} with the user document
*there is no performance problem here, it will be max 1 project i guess with that _id , you only do 1 test to see if its already contributor.
Test code here
update(
{"_id": {"$eq": 1}}, //find the project
[{"$set":
{"contributors":
{"$cond":
[{"$in": [2, "$contributors._id"]}, "$contributors",
{"$concatArrays": ["$contributors", [{"_id": 2}]]}]}}}])
you need to use $addToSet when updating your project document .
$addToSet will only add element to array if it is not added .
let user_id = req.body.user_id;
let project_id = req.body.project_id;
let data_to_update = { $addToSet:{contributors: user_id}};
Project.updateOne({ _id:project_id },data_to_update).then((updated)=>{
console.log('updated =',updated);
return res.status(200).json(updated);
}).catch((err)=>{
});

Get some objects from array of objects in a document (mongoose)

I want to get some objects from array of objects in a mongodb document
my document looks something like this
{
movieDetails: {
movieId: 333,
movieName: "movie",
},
Dates: [
{
Date: " 01/07/2021",
TheaterId: 12,
Is3D: "false",
},
{
Date: " 09/07/2021",
TheaterId: 13,
Is3D: "false",
},
{
Date: " 03/07/2021",
TheaterId: 12,
Is3D: "false",
}
]
}
I want to get only the objects where the movieId is equal to 333 and TheaterId is 12 my result should look like this
[
{
Date: " 01/07/2021",
TheaterId: 12,
Is3D: "false",
},
{
Date: " 03/07/2021",
TheaterId: 12,
Is3D: "false",
}
]
I tried this
const dates = await Movie.find("movieDetails.movieId": 333).select({ Dates: {$elemMatch: {TheaterId: 12}}});
But it returns only the first object if anyone can help me with this I will be very grateful
According to $elemMatch docs:
The $elemMatch operator limits the contents of an field from the query results to contain only the first element matching the $elemMatch condition.
So you can use this aggregation pipeline using $filter into a $project stage like this:
Movie.aggregate([
{
"$match": {
"movieDetails.movieId": 333
}
},
{
"$project": {
"movieDetails": 1,
"Dates": {
"$filter": {
"input": "$Dates",
"as": "d",
"cond": {
"$eq": [
"$$d.TheaterId",
12
]
}
}
}
}
}
])
Example here
Or this another pipeline (I prefer using $filter)
Movie.aggregate([
{
"$match": {
"movieDetails.movieId": 333
}
},
{
"$unwind": "$Dates"
},
{
"$match": {
"Dates.TheaterId": 12
}
},
{
"$group": {
"_id": "$_id",
"movieDetails": {
"$first": "$movieDetails"
},
"Dates": {
"$push": "$Dates"
}
}
}
])
Example here
I would suggest taking a look at the doc for .find() since it seems you are using it incorrectly.
One way you can achieve this is to use .find() to find the movie with the ID you want, then .filter() the Dates from that object to return just the ones with your TheaterId
here is an example
const movies = [{
movieDetails: {
movieId: 333,
movieName: "movie",
},
Dates: [{
Date: " 01/07/2021",
TheaterId: 12,
Is3D: "false",
},
{
Date: " 09/07/2021",
TheaterId: 13,
Is3D: "false",
},
{
Date: " 03/07/2021",
TheaterId: 12,
Is3D: "false",
}
]
}]
function findMovieTimesByTheater(movieID, theaterID) {
const movie = movies.find(movie => movie.movieDetails.movieId === movieID);
if (typeof movie === "undefined") {
console.log(`movieID: ${movieID} not found`);
return [];
}
return movie.Dates.filter(date => date.TheaterId === theaterID);
}
const times = findMovieTimesByTheater(333, 12);
console.log(times);

Is there a way to count distinct values in mongoose with specific cut off date

For example, I have the following two docs
[
{
"_id": "5fc534505144dd0030c44f8e",
"createdAt": "2020-12-14T15:11:21.327Z"
"user_id": "2",
},
{
"_id": "5fc534505144dd0030c44f8e",
"createdAt": "2020-12-14T14:10:40.427Z",
"user_id": "1"
},
{
"_id": "5fc534595144dd0030c44f95",
"createdAt": "2020-12-13T14:10:58.027Z",
"user_id": "1"
}
]
the results should be
[
{
"date": "2020-12-13",
"count":1
},
{
"date": "2020-12-14",
"count":2
}
]
where the count is the number of distinct docs via user_ids till the date that specific cut off date
given data
data=[
{
"createdAt": "2020-12-14T15:11:21.327Z",
"user_id": "2",
},
{
"createdAt": "2020-12-14T14:10:40.427Z",
"user_id": "1"
},
{
"createdAt": "2020-12-13T14:10:58.027Z",
"user_id": "1"
},{
"createdAt": new Date("2020-12-14T14:10:58.027Z"),
}
]
> db.dummy.insert(data)
You may use aggregate: use $group with _id being the date's day in conjunction with $sum)
> db.dummy.aggregate({$group:{_id:{$substr:['$createdAt', 0, 10]}, count:{$sum:1}}})
{ "_id" : "2020-12-14", "count" : 3 }
{ "_id" : "2020-12-13", "count" : 1 }
edit: mongoose wise same may hold
const mongoose = require('mongoose')
mongoose.connect('mongodb://localhost:27017/dummy')
const UDate = mongoose.model('X', { createdAt:String, user_id: String }, 'dummy')
;(async()=>{
mongoose.set('debug', true)
const group = {$group:{_id:{$substr:['$createdAt', 0, 10]}, count:{$sum:1}}}
const v = await UDate.aggregate([group])
console.log('s : ', JSON.stringify(v))
mongoose.disconnect()
})()
edit2: to handle unicity of userIds so there are not counted twice per date, you may use $addToSet instead of sum followed by a projection using $size
const group = {$group:{_id:{$substr:['$createdAt', 0, 10]}, userIds:{$addToSet:'$user_id'}}}
const count = {$project:{date:'$_id', count: {$size: '$userIds'}} }
const v = await Stock.aggregate([group, count])
Lastly, if you feel always more, you can "rename" the _id field as date during the projection
{$project:{date:'$_id', _id:0, count: {$size: '$userIds'}} }
$gorup by createdAt date after getting substring using $substr and make array of unique user ids on the base of $addToset
get total element in count array using $size
db.collection.aggregate([
{
$group: {
_id: { $substr: ["$createdAt", 0, 10] },
count: { $addToSet: "$user_id" }
}
},
{ $addFields: { count: { $size: "$count" } } }
])
Playground

How to query an array in mongodb

Was trying to filter an array with another condition to query my MongoDB database
I have tried using the elemMatch to match exactly with the query, but it not working out.
Here is my code
my shipment schema
const mongoose = require('mongoose');
const Schema = mongoose.Schema;
// Create Schema
const ShipmentSchema = new Schema({
warehouseNo:{
type: String,
ref: 'users.unitNo'
},
packages:[
{
category:{
type: String
},
quantity:{
type: String
},
description:{
type: String
},
trackingno:{
type: String,
},
date:{
type: Date,
default: Date.now
},
length:{
type: Number
},
width:{
type: Number
},
height:{
type: Number
},
weight:{
type: Number
},
fee:{
type: Number,
},
status: {
type: String,
default: "In warehouse"
},
},
],
shippingMode:{
type: String,
},
date:{
type: Date,
default: Date.now
}
});
module.exports = Shipments = mongoose.model('shipments', ShipmentSchema);
Here is my node js.
// #route GET api/user/package
// #desc Get all package
// #access Private
router.get('/package',
passport.authenticate('jwt', { session: false }),
(req, res) => {
const errors = {};
Shipments.findOne({warehouseNo : req.user.unitNo})
.then(shipments => {
if (shipments.packages.length === 0) {
errors.nopackages = 'There are no packages for you yet';
return res.status(404).json(errors);
}
res.json(shipments.packages);
})
});
The code above bring every record in my mongoddb, but if i tried the below, where i ask it to fillter by package status. i got a code crash error
// #route GET api/user/package
// #desc Get all package
// #access Private
router.get('/package',
passport.authenticate('jwt', { session: false }),
(req, res) => {
const errors = {};
Shipments.find({warehouseNo : req.user.unitNo, "packages.status": "In warehouse"})
.then(shipments => {
if (shipments.packages.length === 0) {
errors.nopackages = 'There are no packages for you yet';
return res.status(404).json(errors);
}
res.json(shipments.packages);
})
});
i expect to get something like this
{
"status": "In warehouse",
"date": "2019-09-11T10:19:02.834Z",
"_id": "5d78ca160e47be29e13253b5",
"category": "liquid",
"quantity": "10 pieces",
"description": "garri",
"trackingno": "MHS085533395",
"weight": 123,
"length": 12,
"height": 12,
"width": 13
}
instead i got this
[
{
"status": "Shipped",
"date": "2019-09-11T10:17:46.485Z",
"_id": "5d78c9ca0e47be29e13253b4",
"category": "liquid",
"quantity": "10 pieces",
"description": "garri",
"trackingno": "SDG561920753",
"weight": 123,
"height": 12,
"width": 13
},
{
"status": "In warehouse",
"date": "2019-09-11T10:19:02.834Z",
"_id": "5d78ca160e47be29e13253b5",
"category": "liquid",
"quantity": "10 pieces",
"description": "garri",
"trackingno": "MHS085533395",
"weight": 123,
"length": 12,
"height": 12,
"width": 13
}
]
You should use $elemMatch inside the key packages, i.e. db.getCollection('shipments').find( {warehouseNo: "123"},
{ packages: { $elemMatch: { status: "In warehouse" }}}).
For Example:
I have a collection as below:
{
"_id" : 1.0,
"name" : {
"first" : "John",
"last" : "Backus"
},
"birth" : ISODate("1924-12-03T05:00:00.000Z"),
"death" : ISODate("2007-03-17T04:00:00.000Z"),
"contribs" : [
"Fortran",
"ALGOL",
"Backus-Naur Form",
"FP"
],
"awards" : [
{
"award" : "W.W. McDowell Award",
"year" : 1967.0,
"by" : "IEEE Computer Society"
},
{
"award" : "National Medal of Science",
"year" : 1975.0,
"by" : "National Science Foundation"
},
{
"award" : "Turing Award",
"year" : 1977.0,
"by" : "ACM"
},
{
"award" : "Draper Prize",
"year" : 1993.0,
"by" : "National Academy of Engineering"
}
]
}
Using query like this:
db.getCollection('bios').find( {_id: 1.0 },
{ awards: { $elemMatch: { year: 1967.0 }}})
Gave me a result:
{
"_id" : 1.0,
"awards" : [
{
"award" : "W.W. McDowell Award",
"year" : 1967.0,
"by" : "IEEE Computer Society"
}
]
}
Hope this will help you.
You defined warehouseNo as reference from other table. It must be some ID. Please make sure you are comparing the same

Many to Many with Mongoose

I have the two models:
Item.js
const mongoose = require('mongoose');
const itemSchema = new mongoose.Schema({
name: String,
stores: [{ type: mongoose.Schema.Types.ObjectId, ref: 'Store' }]
});
const Item = mongoose.model('Item', itemSchema);
module.exports = Item;
Store.js
const mongoose = require('mongoose');
const storeSchema = new mongoose.Schema({
name: String,
items: [{ type: mongoose.Schema.Types.ObjectId, ref: 'Item' }]
});
const Store = mongoose.model('Store', storeSchema);
module.exports = Store;
And a seed.js file:
const faker = require('faker');
const Store = require('./models/Store');
const Item = require('./models/Item');
console.log('Seeding..');
let item = new Item({
name: faker.name.findName() + " Item"
});
item.save((err) => {
if (err) return;
let store = new Store({
name: faker.name.findName() + " Store"
});
store.items.push(item);
store.save((err) => {
if (err) return;
})
});
The store is saved with the items array containing 1 item. The item though, doesn't have stores. What am I missing? How to automatically update the many-to-many relationships in MongoDB/Mongoose? I was used to Rails and everything was done automatically.
The problem you presently have is that you saved the reference in one model but you did not save it in the other. There is no "automatic referential integrity" in MongoDB, and such concept of "relations" are really a "manual" affair, and in fact the case with .populate() is actually a whole bunch of additional queries in order to retrieve the referenced information. No "magic" here.
Correct handling of "many to many" comes down to three options:
Listing 1 - Keep arrays on Both documents
Following your current design, the parts you are missing is storing the referenced on "both" the related items. For a listing to demonstrate:
const { Schema } = mongoose = require('mongoose');
mongoose.Promise = global.Promise;
mongoose.set('debug',true);
mongoose.set('useFindAndModify', false);
mongoose.set('useCreateIndex', true);
const uri = 'mongodb://localhost:27017/manydemo',
options = { useNewUrlParser: true };
const itemSchema = new Schema({
name: String,
stores: [{ type: Schema.Types.ObjectId, ref: 'Store' }]
});
const storeSchema = new Schema({
name: String,
items: [{ type: Schema.Types.ObjectId, ref: 'Item' }]
});
const Item = mongoose.model('Item', itemSchema);
const Store = mongoose.model('Store', storeSchema);
const log = data => console.log(JSON.stringify(data,undefined,2))
(async function() {
try {
const conn = await mongoose.connect(uri,options);
// Clean data
await Promise.all(
Object.entries(conn.models).map(([k,m]) => m.deleteMany() )
);
// Create some instances
let [toothpaste,brush] = ['toothpaste','brush'].map(
name => new Item({ name })
);
let [billsStore,tedsStore] = ['Bills','Teds'].map(
name => new Store({ name })
);
// Add items to stores
[billsStore,tedsStore].forEach( store => {
store.items.push(toothpaste); // add toothpaste to store
toothpaste.stores.push(store); // add store to toothpaste
});
// Brush is only in billsStore
billsStore.items.push(brush);
brush.stores.push(billsStore);
// Save everything
await Promise.all(
[toothpaste,brush,billsStore,tedsStore].map( m => m.save() )
);
// Show stores
let stores = await Store.find().populate('items','-stores');
log(stores);
// Show items
let items = await Item.find().populate('stores','-items');
log(items);
} catch(e) {
console.error(e);
} finally {
mongoose.disconnect();
}
})();
This creates the "items" collection:
{
"_id" : ObjectId("59ab96d9c079220dd8eec428"),
"name" : "toothpaste",
"stores" : [
ObjectId("59ab96d9c079220dd8eec42a"),
ObjectId("59ab96d9c079220dd8eec42b")
],
"__v" : 0
}
{
"_id" : ObjectId("59ab96d9c079220dd8eec429"),
"name" : "brush",
"stores" : [
ObjectId("59ab96d9c079220dd8eec42a")
],
"__v" : 0
}
And the "stores" collection:
{
"_id" : ObjectId("59ab96d9c079220dd8eec42a"),
"name" : "Bills",
"items" : [
ObjectId("59ab96d9c079220dd8eec428"),
ObjectId("59ab96d9c079220dd8eec429")
],
"__v" : 0
}
{
"_id" : ObjectId("59ab96d9c079220dd8eec42b"),
"name" : "Teds",
"items" : [
ObjectId("59ab96d9c079220dd8eec428")
],
"__v" : 0
}
And produces overall output such as:
Mongoose: items.deleteMany({}, {})
Mongoose: stores.deleteMany({}, {})
Mongoose: items.insertOne({ name: 'toothpaste', _id: ObjectId("59ab96d9c079220dd8eec428"), stores: [ ObjectId("59ab96d9c079220dd8eec42a"), ObjectId("59ab96d9c079220dd8eec42b") ], __v: 0 })
Mongoose: items.insertOne({ name: 'brush', _id: ObjectId("59ab96d9c079220dd8eec429"), stores: [ ObjectId("59ab96d9c079220dd8eec42a") ], __v: 0 })
Mongoose: stores.insertOne({ name: 'Bills', _id: ObjectId("59ab96d9c079220dd8eec42a"), items: [ ObjectId("59ab96d9c079220dd8eec428"), ObjectId("59ab96d9c079220dd8eec429") ], __v: 0 })
Mongoose: stores.insertOne({ name: 'Teds', _id: ObjectId("59ab96d9c079220dd8eec42b"), items: [ ObjectId("59ab96d9c079220dd8eec428") ], __v: 0 })
Mongoose: stores.find({}, { fields: {} })
Mongoose: items.find({ _id: { '$in': [ ObjectId("59ab96d9c079220dd8eec428"), ObjectId("59ab96d9c079220dd8eec429") ] } }, { fields: { stores: 0 } })
[
{
"_id": "59ab96d9c079220dd8eec42a",
"name": "Bills",
"__v": 0,
"items": [
{
"_id": "59ab96d9c079220dd8eec428",
"name": "toothpaste",
"__v": 0
},
{
"_id": "59ab96d9c079220dd8eec429",
"name": "brush",
"__v": 0
}
]
},
{
"_id": "59ab96d9c079220dd8eec42b",
"name": "Teds",
"__v": 0,
"items": [
{
"_id": "59ab96d9c079220dd8eec428",
"name": "toothpaste",
"__v": 0
}
]
}
]
Mongoose: items.find({}, { fields: {} })
Mongoose: stores.find({ _id: { '$in': [ ObjectId("59ab96d9c079220dd8eec42a"), ObjectId("59ab96d9c079220dd8eec42b") ] } }, { fields: { items: 0 } })
[
{
"_id": "59ab96d9c079220dd8eec428",
"name": "toothpaste",
"__v": 0,
"stores": [
{
"_id": "59ab96d9c079220dd8eec42a",
"name": "Bills",
"__v": 0
},
{
"_id": "59ab96d9c079220dd8eec42b",
"name": "Teds",
"__v": 0
}
]
},
{
"_id": "59ab96d9c079220dd8eec429",
"name": "brush",
"__v": 0,
"stores": [
{
"_id": "59ab96d9c079220dd8eec42a",
"name": "Bills",
"__v": 0
}
]
}
]
The key points being that you actually add the reference data to each document in each collection where a relationship exists. The "arrays" present are used here to store those references and "lookup" the results from the related collection and replace them with the object data that was stored there.
Pay attention to parts like:
// Add items to stores
[billsStore,tedsStore].forEach( store => {
store.items.push(toothpaste); // add toothpaste to store
toothpaste.stores.push(store); // add store to toothpaste
});
Because that means not only are we adding the toothpaste to the "items" array in each store, but we are also adding each "store" to the "stores" array of the toothpaste item. This is done so the relationships can work being queried from either direction. If you only wanted "items from stores" and never "stores from items", then you would not need to store the relation data on the "item" entries at all.
Listing 2 - Use Virtuals and an Intermediary Collection
This is essentially the classic "many to many" relation. Where instead of directly defining relationships between the two collections, there is another collection ( table ) that stores the details about which item is related to which store.
As a full listing:
const { Schema } = mongoose = require('mongoose');
mongoose.Promise = global.Promise;
mongoose.set('debug',true);
mongoose.set('useFindAndModify', false);
mongoose.set('useCreateIndex', true);
const uri = 'mongodb://localhost:27017/manydemo',
options = { useNewUrlParser: true };
const itemSchema = new Schema({
name: String,
},{
toJSON: { virtuals: true }
});
itemSchema.virtual('stores', {
ref: 'StoreItem',
localField: '_id',
foreignField: 'itemId'
});
const storeSchema = new Schema({
name: String,
},{
toJSON: { virtuals: true }
});
storeSchema.virtual('items', {
ref: 'StoreItem',
localField: '_id',
foreignField: 'storeId'
});
const storeItemSchema = new Schema({
storeId: { type: Schema.Types.ObjectId, ref: 'Store', required: true },
itemId: { type: Schema.Types.ObjectId, ref: 'Item', required: true }
});
const Item = mongoose.model('Item', itemSchema);
const Store = mongoose.model('Store', storeSchema);
const StoreItem = mongoose.model('StoreItem', storeItemSchema);
const log = data => console.log(JSON.stringify(data,undefined,2));
(async function() {
try {
const conn = await mongoose.connect(uri,options);
// Clean data
await Promise.all(
Object.entries(conn.models).map(([k,m]) => m.deleteMany() )
);
// Create some instances
let [toothpaste,brush] = await Item.insertMany(
['toothpaste','brush'].map( name => ({ name }) )
);
let [billsStore,tedsStore] = await Store.insertMany(
['Bills','Teds'].map( name => ({ name }) )
);
// Add toothpaste to both stores
for( let store of [billsStore,tedsStore] ) {
await StoreItem.update(
{ storeId: store._id, itemId: toothpaste._id },
{ },
{ 'upsert': true }
);
}
// Add brush to billsStore
await StoreItem.update(
{ storeId: billsStore._id, itemId: brush._id },
{},
{ 'upsert': true }
);
// Show stores
let stores = await Store.find().populate({
path: 'items',
populate: { path: 'itemId' }
});
log(stores);
// Show Items
let items = await Item.find().populate({
path: 'stores',
populate: { path: 'storeId' }
});
log(items);
} catch(e) {
console.error(e);
} finally {
mongoose.disconnect();
}
})();
The relations are now in their own collection, so the data now appears differently, for "items":
{
"_id" : ObjectId("59ab996166d5cc0e0d164d74"),
"__v" : 0,
"name" : "toothpaste"
}
{
"_id" : ObjectId("59ab996166d5cc0e0d164d75"),
"__v" : 0,
"name" : "brush"
}
And "stores":
{
"_id" : ObjectId("59ab996166d5cc0e0d164d76"),
"__v" : 0,
"name" : "Bills"
}
{
"_id" : ObjectId("59ab996166d5cc0e0d164d77"),
"__v" : 0,
"name" : "Teds"
}
And now for "storeitems" which maps the relations:
{
"_id" : ObjectId("59ab996179e41cc54405b72b"),
"itemId" : ObjectId("59ab996166d5cc0e0d164d74"),
"storeId" : ObjectId("59ab996166d5cc0e0d164d76"),
"__v" : 0
}
{
"_id" : ObjectId("59ab996179e41cc54405b72d"),
"itemId" : ObjectId("59ab996166d5cc0e0d164d74"),
"storeId" : ObjectId("59ab996166d5cc0e0d164d77"),
"__v" : 0
}
{
"_id" : ObjectId("59ab996179e41cc54405b72f"),
"itemId" : ObjectId("59ab996166d5cc0e0d164d75"),
"storeId" : ObjectId("59ab996166d5cc0e0d164d76"),
"__v" : 0
}
With full output like:
Mongoose: items.deleteMany({}, {})
Mongoose: stores.deleteMany({}, {})
Mongoose: storeitems.deleteMany({}, {})
Mongoose: items.insertMany([ { __v: 0, name: 'toothpaste', _id: 59ab996166d5cc0e0d164d74 }, { __v: 0, name: 'brush', _id: 59ab996166d5cc0e0d164d75 } ])
Mongoose: stores.insertMany([ { __v: 0, name: 'Bills', _id: 59ab996166d5cc0e0d164d76 }, { __v: 0, name: 'Teds', _id: 59ab996166d5cc0e0d164d77 } ])
Mongoose: storeitems.update({ itemId: ObjectId("59ab996166d5cc0e0d164d74"), storeId: ObjectId("59ab996166d5cc0e0d164d76") }, { '$setOnInsert': { __v: 0 } }, { upsert: true })
Mongoose: storeitems.update({ itemId: ObjectId("59ab996166d5cc0e0d164d74"), storeId: ObjectId("59ab996166d5cc0e0d164d77") }, { '$setOnInsert': { __v: 0 } }, { upsert: true })
Mongoose: storeitems.update({ itemId: ObjectId("59ab996166d5cc0e0d164d75"), storeId: ObjectId("59ab996166d5cc0e0d164d76") }, { '$setOnInsert': { __v: 0 } }, { upsert: true })
Mongoose: stores.find({}, { fields: {} })
Mongoose: storeitems.find({ storeId: { '$in': [ ObjectId("59ab996166d5cc0e0d164d76"), ObjectId("59ab996166d5cc0e0d164d77") ] } }, { fields: {} })
Mongoose: items.find({ _id: { '$in': [ ObjectId("59ab996166d5cc0e0d164d74"), ObjectId("59ab996166d5cc0e0d164d75") ] } }, { fields: {} })
[
{
"_id": "59ab996166d5cc0e0d164d76",
"__v": 0,
"name": "Bills",
"items": [
{
"_id": "59ab996179e41cc54405b72b",
"itemId": {
"_id": "59ab996166d5cc0e0d164d74",
"__v": 0,
"name": "toothpaste",
"stores": null,
"id": "59ab996166d5cc0e0d164d74"
},
"storeId": "59ab996166d5cc0e0d164d76",
"__v": 0
},
{
"_id": "59ab996179e41cc54405b72f",
"itemId": {
"_id": "59ab996166d5cc0e0d164d75",
"__v": 0,
"name": "brush",
"stores": null,
"id": "59ab996166d5cc0e0d164d75"
},
"storeId": "59ab996166d5cc0e0d164d76",
"__v": 0
}
],
"id": "59ab996166d5cc0e0d164d76"
},
{
"_id": "59ab996166d5cc0e0d164d77",
"__v": 0,
"name": "Teds",
"items": [
{
"_id": "59ab996179e41cc54405b72d",
"itemId": {
"_id": "59ab996166d5cc0e0d164d74",
"__v": 0,
"name": "toothpaste",
"stores": null,
"id": "59ab996166d5cc0e0d164d74"
},
"storeId": "59ab996166d5cc0e0d164d77",
"__v": 0
}
],
"id": "59ab996166d5cc0e0d164d77"
}
]
Mongoose: items.find({}, { fields: {} })
Mongoose: storeitems.find({ itemId: { '$in': [ ObjectId("59ab996166d5cc0e0d164d74"), ObjectId("59ab996166d5cc0e0d164d75") ] } }, { fields: {} })
Mongoose: stores.find({ _id: { '$in': [ ObjectId("59ab996166d5cc0e0d164d76"), ObjectId("59ab996166d5cc0e0d164d77") ] } }, { fields: {} })
[
{
"_id": "59ab996166d5cc0e0d164d74",
"__v": 0,
"name": "toothpaste",
"stores": [
{
"_id": "59ab996179e41cc54405b72b",
"itemId": "59ab996166d5cc0e0d164d74",
"storeId": {
"_id": "59ab996166d5cc0e0d164d76",
"__v": 0,
"name": "Bills",
"items": null,
"id": "59ab996166d5cc0e0d164d76"
},
"__v": 0
},
{
"_id": "59ab996179e41cc54405b72d",
"itemId": "59ab996166d5cc0e0d164d74",
"storeId": {
"_id": "59ab996166d5cc0e0d164d77",
"__v": 0,
"name": "Teds",
"items": null,
"id": "59ab996166d5cc0e0d164d77"
},
"__v": 0
}
],
"id": "59ab996166d5cc0e0d164d74"
},
{
"_id": "59ab996166d5cc0e0d164d75",
"__v": 0,
"name": "brush",
"stores": [
{
"_id": "59ab996179e41cc54405b72f",
"itemId": "59ab996166d5cc0e0d164d75",
"storeId": {
"_id": "59ab996166d5cc0e0d164d76",
"__v": 0,
"name": "Bills",
"items": null,
"id": "59ab996166d5cc0e0d164d76"
},
"__v": 0
}
],
"id": "59ab996166d5cc0e0d164d75"
}
]
Since the relations are now mapped in a separate collection there are a couple of changes here. Notably we want to define a "virtual" field on the collection that no longer has a fixed array of items. So you add one as is shown:
const itemSchema = new Schema({
name: String,
},{
toJSON: { virtuals: true }
});
itemSchema.virtual('stores', {
ref: 'StoreItem',
localField: '_id',
foreignField: 'itemId'
});
You assign the virtual field with it's localField and foreignField mappings so the subsequent .populate() call knows what to use.
The intermediary collection has a fairly standard definition:
const storeItemSchema = new Schema({
storeId: { type: Schema.Types.ObjectId, ref: 'Store', required: true },
itemId: { type: Schema.Types.ObjectId, ref: 'Item', required: true }
});
And instead of "pushing" new items onto arrays, we instead add them to this new collection. A reasonable method for this is using "upserts" to create a new entry only when this combination does not exist:
// Add toothpaste to both stores
for( let store of [billsStore,tedsStore] ) {
await StoreItem.update(
{ storeId: store._id, itemId: toothpaste._id },
{ },
{ 'upsert': true }
);
}
It's a pretty simple method that merely creates a new document with the two keys supplied in the query where one was not found, or essentially tries to update the same document when matched, and with "nothing" in this case. So existing matches just end up as a "no-op", which is the desired thing to do. Alternately you could simply .insertOne() an ignore duplicate key errors. Whatever takes your fancy.
Actually querying this "related" data works a little differently again. Because there is another collection involved, we call .populate() in a way that considers it needs to "lookup" the relation on other retrieved property as well. So you have calls like this:
// Show stores
let stores = await Store.find().populate({
path: 'items',
populate: { path: 'itemId' }
});
log(stores);
Listing 3 - Use Modern Features to do it on the server
So depending on which approach taken, being using arrays or an intermediary collection to store the relation data in as an alternative to "growing arrays" within the documents, then the obvious thing you should be noting is that the .populate() calls used are actually making additional queries to MongoDB and pulling those documents over the network in separate requests.
This might appear all well and fine in small doses, however as things scale up and especially over volumes of requests, this is never a good thing. Additionally there might well be other conditions you want to apply that means you don't need to pull all the documents from the server, and would rather match data from those "relations" before you returned results.
This is why modern MongoDB releases include $lookup which actually "joins" the data on the server itself. By now you should have been looking at all the output those API calls produce as shown by mongoose.set('debug',true).
So instead of producing multiple queries, this time we make it one aggregation statement to "join" on the server, and return the results in one request:
// Show Stores
let stores = await Store.aggregate([
{ '$lookup': {
'from': StoreItem.collection.name,
'let': { 'id': '$_id' },
'pipeline': [
{ '$match': {
'$expr': { '$eq': [ '$$id', '$storeId' ] }
}},
{ '$lookup': {
'from': Item.collection.name,
'let': { 'itemId': '$itemId' },
'pipeline': [
{ '$match': {
'$expr': { '$eq': [ '$_id', '$$itemId' ] }
}}
],
'as': 'items'
}},
{ '$unwind': '$items' },
{ '$replaceRoot': { 'newRoot': '$items' } }
],
'as': 'items'
}}
])
log(stores);
Which whilst longer in coding, is actually far superior in efficiency even for the very trivial action right here. This of course scales considerably.
Following the same "intermediary" model as before ( and just for example, because it could be done either way ) we have a full listing:
const { Schema } = mongoose = require('mongoose');
const uri = 'mongodb://localhost:27017/manydemo',
options = { useNewUrlParser: true };
mongoose.Promise = global.Promise;
mongoose.set('debug', true);
mongoose.set('useFindAndModify', false);
mongoose.set('useCreateIndex', true);
const itemSchema = new Schema({
name: String
}, {
toJSON: { virtuals: true }
});
itemSchema.virtual('stores', {
ref: 'StoreItem',
localField: '_id',
foreignField: 'itemId'
});
const storeSchema = new Schema({
name: String
}, {
toJSON: { virtuals: true }
});
storeSchema.virtual('items', {
ref: 'StoreItem',
localField: '_id',
foreignField: 'storeId'
});
const storeItemSchema = new Schema({
storeId: { type: Schema.Types.ObjectId, ref: 'Store', required: true },
itemId: { type: Schema.Types.ObjectId, ref: 'Item', required: true }
});
const Item = mongoose.model('Item', itemSchema);
const Store = mongoose.model('Store', storeSchema);
const StoreItem = mongoose.model('StoreItem', storeItemSchema);
const log = data => console.log(JSON.stringify(data, undefined, 2));
(async function() {
try {
const conn = await mongoose.connect(uri, options);
// Clean data
await Promise.all(
Object.entries(conn.models).map(([k,m]) => m.deleteMany())
);
// Create some instances
let [toothpaste, brush] = await Item.insertMany(
['toothpaste', 'brush'].map(name => ({ name }) )
);
let [billsStore, tedsStore] = await Store.insertMany(
['Bills', 'Teds'].map( name => ({ name }) )
);
// Add toothpaste to both stores
for ( let { _id: storeId } of [billsStore, tedsStore] ) {
await StoreItem.updateOne(
{ storeId, itemId: toothpaste._id },
{ },
{ 'upsert': true }
);
}
// Add brush to billsStore
await StoreItem.updateOne(
{ storeId: billsStore._id, itemId: brush._id },
{ },
{ 'upsert': true }
);
// Show Stores
let stores = await Store.aggregate([
{ '$lookup': {
'from': StoreItem.collection.name,
'let': { 'id': '$_id' },
'pipeline': [
{ '$match': {
'$expr': { '$eq': [ '$$id', '$storeId' ] }
}},
{ '$lookup': {
'from': Item.collection.name,
'let': { 'itemId': '$itemId' },
'pipeline': [
{ '$match': {
'$expr': { '$eq': [ '$_id', '$$itemId' ] }
}}
],
'as': 'items'
}},
{ '$unwind': '$items' },
{ '$replaceRoot': { 'newRoot': '$items' } }
],
'as': 'items'
}}
])
log(stores);
// Show Items
let items = await Item.aggregate([
{ '$lookup': {
'from': StoreItem.collection.name,
'let': { 'id': '$_id' },
'pipeline': [
{ '$match': {
'$expr': { '$eq': [ '$$id', '$itemId' ] }
}},
{ '$lookup': {
'from': Store.collection.name,
'let': { 'storeId': '$storeId' },
'pipeline': [
{ '$match': {
'$expr': { '$eq': [ '$_id', '$$storeId' ] }
}}
],
'as': 'stores',
}},
{ '$unwind': '$stores' },
{ '$replaceRoot': { 'newRoot': '$stores' } }
],
'as': 'stores'
}}
]);
log(items);
} catch(e) {
console.error(e);
} finally {
mongoose.disconnect();
}
})()
And the output:
Mongoose: stores.aggregate([ { '$lookup': { from: 'storeitems', let: { id: '$_id' }, pipeline: [ { '$match': { '$expr': { '$eq': [ '$$id', '$storeId' ] } } }, { '$lookup': { from: 'items', let: { itemId: '$itemId' }, pipeline: [ { '$match': { '$expr': { '$eq': [ '$_id', '$$itemId' ] } } } ], as: 'items' } }, { '$unwind': '$items' }, { '$replaceRoot': { newRoot: '$items' } } ], as: 'items' } } ], {})
[
{
"_id": "5ca7210717dadc69652b37da",
"name": "Bills",
"__v": 0,
"items": [
{
"_id": "5ca7210717dadc69652b37d8",
"name": "toothpaste",
"__v": 0
},
{
"_id": "5ca7210717dadc69652b37d9",
"name": "brush",
"__v": 0
}
]
},
{
"_id": "5ca7210717dadc69652b37db",
"name": "Teds",
"__v": 0,
"items": [
{
"_id": "5ca7210717dadc69652b37d8",
"name": "toothpaste",
"__v": 0
}
]
}
]
Mongoose: items.aggregate([ { '$lookup': { from: 'storeitems', let: { id: '$_id' }, pipeline: [ { '$match': { '$expr': { '$eq': [ '$$id', '$itemId' ] } } }, { '$lookup': { from: 'stores', let: { storeId: '$storeId' }, pipeline: [ { '$match': { '$expr': { '$eq': [ '$_id', '$$storeId' ] } } } ], as: 'stores' } }, { '$unwind': '$stores' }, { '$replaceRoot': { newRoot: '$stores' } } ], as: 'stores' } } ], {})
[
{
"_id": "5ca7210717dadc69652b37d8",
"name": "toothpaste",
"__v": 0,
"stores": [
{
"_id": "5ca7210717dadc69652b37da",
"name": "Bills",
"__v": 0
},
{
"_id": "5ca7210717dadc69652b37db",
"name": "Teds",
"__v": 0
}
]
},
{
"_id": "5ca7210717dadc69652b37d9",
"name": "brush",
"__v": 0,
"stores": [
{
"_id": "5ca7210717dadc69652b37da",
"name": "Bills",
"__v": 0
}
]
}
]
What should be obvious is the significant reduction in the queries issued on the end to return the "joined" form of the data. This means lower latency and more responsive applications as a result of removing all the network overhead.
Final Notes
Those a are generally your approaches to dealing with "many to many" relations, which essentially comes down to either:
Keeping arrays in each document on either side holding the references to the related items.
Storing an intermediary collection and using that as a lookup reference to retrieving the other data.
In all cases it is up to you to actually store those references if you expect things to work on "both directions". Of course $lookup and even "virtuals" where that applies means that you don't always need to store on every source since you could then "reference" in just one place and use that information by applying those methods.
The other case is of course "embedding", which is an entirely different game and what document oriented databases such as MongoDB are really all about. Therefore instead of "fetching from another collection" the concept is of course to "embed" the data.
This means not just the ObjectId values that point to the other items, but actually storing the full data within arrays in each document. There is of course an issue of "size" and of course issues with updating data in multiple places. This is generally the trade off for there being a single request and a simple request that does not need to go and find data in other collections because it's "already there".
There is plenty of material around on the subject of referencing vs embedding. Once such summary source is Mongoose populate vs object nesting or even the very general MongoDB relationships: embed or reference? and many many others.
You should spend some time thinking about the concepts and how this applies to your application in general. And note that you are not actually using an RDBMS here, so you might as well use the correct features that you are meant to exploit, rather than simply making one act like the other.
You first should consider the usage of data in your application before modeling the database.
I don't have the detailed requirements of your application. But why do you have to keep 2 references in 2 schemas? Why not just keep 1 reference from Store to Item (which means 1 store has many items), and then if you want execute a query to find which stores does a item belong to, there is also away to do it by querying Store collection.
In addition, there is nothing called "many-to-many" in MongoDB. It depends on how the data is being used that you must figure out the efficient way to form the relationship between collections, as well as to structure your database.
Anyway, if you still want to use your current schemas, you can first create the item, then create the store and push the created item's id in to items array, then execute a update to the item with created store's id.

Categories

Resources