How to group the documents which I just unwinded in mongodb? - mongodb

Question

This is the initial document. I applied $unwind on "colors" and another $unwind on "colors.sizes". But now am not able to group them back. How is that done?
{
"_id": ObjectId("5ef838255c959771c46fc917"),
"created": ISODate("2017-03-04T18:30:00.000Z"),
"updated": ISODate("2020-06-28T00:49:23.000Z"),
"status": "active",
"productId": "3828",
"tags": {
"colors": ["BEIGE", "BLACK", "BLUE", "GREEN", "GREY", "LIGHT BEIGE", "LIGHT PINK", "NAVY", "PINK", "WHITE"],
"size": ["39", "39.5", "40", "40.5", "41", "41.5", "42", "42.5", "43", "43.5", "44", "44.5", "45", "45.5", "46"],
"category": ["SHOES", "SNEAKERS", "MENS SHOES & ACCESSORIES"],
"price": ["3400.00", "1360.00", "2040.00", "1700.00", "45575.00"],
"season": ["SS(2020)", "AW(2019)", "SS(2019)", "AW(2018)", "SS(2017)", "AW(2016)", "SS(2016)", "SS(2015)"],
"hash": ["ACTIVE", "ANKLE", "BLUSH", "COLLAR", "COLOUR", "DESIGNER", "FOIL", "GOLD-TONE", "LACE-UP", "LEATHER", "LOW", "LOW TOP", "ORIGINAL", "PADDED", "ROUND", "ROUND TOE", "RUBBER", "SIGNATURE", "TONAL"]
},
"name": "ORIGINAL ARCHILLES LOW SNEAKER",
"description": "",
"brand": "COMMON PROJECTS",
"styleCode": "220036200",
"availableOnline": true,
"colors": [{
"images": [],
"_id": ObjectId("5ef838255c959771c46fc918"),
"colorId": "1",
"color": "WHITE",
"hexCode": "#FFFFFF",
"status": "active",
"sizes": [{
"extraInfo": [{
"title": "Size And Fit",
"text": "Product measures: Heel 3cm (1.2\"). True to size"
}, {
"title": "Information",
"text": "Designer colour: White. Leather. Rubber outsole. Made in Italy"
}],
"_id": ObjectId("5ef838255c959771c46fc919"),
"sizeId": "1",
"neo": "0220003258813",
"size": "39",
"originalPrice": "3400.00",
"sellingPrice": "3400.00",
"discountPercent": "0.00",
"status": "active"
}, {
"extraInfo": [{
"title": "Size And Fit",
"text": "Product measures: Heel 3cm (1.2\"). True to size"
}, {
"title": "Information",
"text": "Designer colour: White. Leather. Rubber outsole. Made in Italy"
}],
"_id": ObjectId("5ef838255c959771c46fc91a"),
"sizeId": "2",
"neo": "0220111220849",
"size": "39.5",
"originalPrice": "3400.00",
"sellingPrice": "3400.00",
"discountPercent": "0.00",
"status": "active"
}, {
"extraInfo": [{
"title": "Size And Fit",
"text": "Product measures: Heel 3cm (1.2\"). True to size"
}, {
"title": "Information",
"text": "Designer colour: White. Leather. Rubber outsole. Made in Italy"
}],
"_id": ObjectId("5ef838255c959771c46fc91b"),
"sizeId": "3",
"neo": "0220004218564",
"size": "40",
"originalPrice": "3400.00",
"sellingPrice": "3400.00",
"discountPercent": "0.00",
"status": "active"
}]
}, ]
}
db.getCollection('products').aggregate([
{$match:{"productId":"3828"}},
{$unwind:"$colors"},
{$unwind:"$colors.sizes"},
{
$lookup: {
from:"inventories",
localField: "colors.sizes.neo",
foreignField: "sku",
as:"colors.sizes.stores"
}
}
])

Answer 1

The opposite operator of $unwind is $group
Explanation
We $group by productId + colors[*]._id fields grouping root, colors and sizes separately.
Note: You can define each variable separately like this:
created: {$first:"$created"},
updated: {$first:"$updated"},
...
OR
use aggregation variable $$ROOT which contains entire document fields
We "merge" each variables into single object and move into root level.
Note: If we don't use $replaceRoot, the data structure will looks like this:
{
_id: "...",
root: {
_id: "...",
...
colors:{
...
sizes:[...]
}
}
}
Note: Unfortunately, MongoDB doesn't allow merge subkeys like this:
{
$mergeObjects:[
"$root",
{"colors":"$colors"},
{"colors.sizes":"$sizes"} <-- Not allowed
]
}
So we have to use $objectToArray + $map + $arrayToObject operators
$objectToArray - transforms object into array
{"hello": "world"} ---> [{"k":"hello", "v":"world"}]
$map - iterates array and transforms array content.
In this case, we replace `sizes` field with `sizes` from the 1st step
$objectToArray - transforms "special" array into object
[{"k":"hello", "v":"world"}] ---> {"hello": "world"}
Now we group only by productId and group colors and root values separately
We finish merging variables and transform into desired object (same as 2nd step)
Add these steps into your pipeline:
{
$group: {
"_id": {
productId: "$productId",
colors: "$colors._id"
},
"root": {
$first: "$$ROOT"
},
"colors": {
$first: "$colors"
},
"sizes": {
$push: "$colors.sizes"
}
}
},
{
$replaceRoot: {
newRoot: {
$mergeObjects: [
"$root",
{
"colors": {
$arrayToObject: {
$map: {
input: { $objectToArray: "$colors" },
as: "color",
in: {
k: "$$color.k",
v: {
$cond: [
{ $eq: [ "$$color.k", "sizes" ] },
"$sizes",
"$$color.v"
]
}
}
}
}
}
}
]
}
}
},
{
$group: {
"_id": "$_id",
"root": {
$first: "$$ROOT"
},
"colors": {
$push: "$colors"
}
}
},
{
$replaceRoot: {
newRoot: {
$mergeObjects: [
"$root",
{ "colors": "$colors" }
]
}
}
}
MongoPlayground