Mongodb loop through every distinct values and select tags using aggregate (facet) - mongodb

Question

I have collection like this:
{
"labels": [{
"description": "Dog"
}, {
"description": "Red"
}, {
"description": "XXX"
}]
}
{
"labels": [{
"description": "Cat"
}, {
"description": "XXX"
}, {
"description": "Yellow"
}]
}
{
"labels": [{
"description": "Dog"
}, {
"description": "Red"
}, {
"description": "Yellow"
}]
}
{
"labels": [{
"description": "Bird"
}, {
"description": "XXX"
}, {
"description": "XXX"
}]
}
I want to filter for example only "Red" and "Yellow" colors from ALL elements and output document like this:
// because "Dog" appears 2 times so total = 2
{
description: "Dog",
total: 2,
colors: [
{ "_id": "Red", total: 2 },
{ "_id": "Yellow", total: 1 }
]
}
{
description: "Cat",
total: 1,
colors: [
{ "_id": "Yellow", total: 1 }
]
}
{
description: "Bird",
total: 1,
colors: []
}
{
description: "Red",
total: 2,
colors: [
{ _id: "Yellow", total: 1 }
]
}
{
description: "XXX",
total: 4,
colors: [
{ _id: "Yellow", total: 1 }
]
}
I can do this by using collection.distinct('labels.description') and then iterating through every single element + make a separate collection.count({ 'labels.description': 'Dog' }) like this:
for (...)
db.collection.aggregate([
{
"$match": {
"labels.description": valueFromLoop // (e.g. Dog)
}
},
{ $unwind : "$labels" },
{
"$group": {
"_id": "$labels.description",
"count": { "$sum": 1 }
}
},
{
"$match": {
"$or": [
{ "_id": "Red" },
{ "_id": "Yellow" }
]
}
},
{
"$sort": {
"count": -1
}
}
])
I want to do this in a single aggregation or mapReduce so that I could easily output it to new collection using $out instead of using Bulk operations separately, however I don't know if it's possible.

Answer 1

Try this:
let filter = ["Red", "Yellow"];
db.testcollection.aggregate([
{
$addFields: { bkp: "$labels" }
},
{ $unwind: "$labels" },
{
$addFields: {
bkp: {
$filter: {
input: "$bkp",
as: "item",
cond: {
$and: [
{ $ne: ["$$item.description", "$labels.description"] },
{ $in: ["$$item.description", filter] }
]
}
}
}
}
},
{
$unwind: {
path: "$bkp",
preserveNullAndEmptyArrays: true
}
},
{
$group: {
_id: {
key1: "$labels.description",
key2: { $ifNull: ["$bkp.description", false] }
},
total: { $sum: 1 }
}
},
{
$group: {
_id: "$_id.key1",
description: { $first: "$_id.key1" },
total: {
$sum: {
$cond: {
if: { $first: [["$_id.key2"]] },
then: 1,
else: "$total"
}
}
},
colors: {
$push: {
$cond: {
if: { $first: [["$_id.key2"]] },
then: {
_id: "$_id.key2",
total: "$total"
},
else: "$$REMOVE"
}
}
}
}
},
{ $project: { _id: 0 } }
]);

Answer 2

For some reason with code from both answers it does not count all tags properly.
I'm posting what works:
db.collection.aggregate([
{
$project: {
labels: 1,
result: {
$filter: {
input: "$labels",
as: "label",
cond: {
$or: [
{ $eq: ["$$label.description", "Blue"] },
{ $eq: ["$$label.description", "Red"] },
{ $eq: ["$$label.description", "Black-and-white"] },
{ $eq: ["$$label.description", "Purple"] },
{ $eq: ["$$label.description", "Orange"] },
{ $eq: ["$$label.description", "Yellow"] },
{ $eq: ["$$label.description", "Green"] },
{ $eq: ["$$label.description", "Teal"] }
]
}
}
}
}
},
{
$unwind: "$labels"
},
{
"$group": {
_id: "$labels.description",
x: {
$push: "$result.description"
},
total: { "$sum": 1 }
}
},
{
$project: {
x: {
$reduce: {
input: '$x',
initialValue: [],
in: {$concatArrays: ['$$value', '$$this']}
}
},
total: 1
}
},
{
$project: {
x: 1,
y: { $setUnion: "$x" },
total: 1
}
},
{
$project: {
_id: 0,
description: "$_id",
"colors": {
$map: {
input: "$y",
as: "item",
in: {
_id: "$$item",
count: {
$size: {
$filter: {
input: "$x",
as: "itemx",
cond: {
$eq: ["$$item", "$$itemx"]
}
}
}
}
}
}
},
total: 1
}
},
{
$out: "backgrounds_meta"
}
])

Answer 3

db.test2.aggregate([
{
$project: {
labels:1,
colours: {
$filter: {
input: "$labels",
as: "label",
cond: {
$or: [
{$eq:["Yellow","$$label.description"]},
{$eq:["Red", "$$label.description"]}
]
}
}
}
}
},
{$unwind:"$labels"},
{$group:{
_id: "$labels.description",
total: {$sum:1},
colours: {$addToSet:"$colours.description"}
}},
{
$project:{
_id:0,
description:"$_id",
total:1,
colours: {
$reduce:{
input: "$colours",
initialValue: [],
in: {$concatArrays: ["$$value", "$$this"]}
}
}
}
},
{
$unwind: {
path:"$colours",preserveNullAndEmptyArrays: true
}
},
{
$group:{
_id:{
description:"$description",
total:"$total",
colour:"$colours"
},
count: {
$sum: {$cond:[{$ifNull:["$colours",false]},1,0]}
}
}
},
{
$group:{
_id:{
description:"$_id.description",
total:"$_id.total"
},
colours: {
$push: {
$cond: [{$gt:["$count",0]},
{
"_id":"$_id.colour",
total:"$count"
},
"$$REMOVE"
]
}
}
}
},
{
$project: {
_id:0,
description: "$_id.description",
total: "$_id.total",
colours: 1
}
}
]);
**Edit In your answer, you are missing the Yellows for Red and Dog because you are taking the first item from $result with $arrayElemAt: ["$result.description", 0].
If description is a colour, do you also want to include the counts for itself in colours?
Never mind, you've updated the answer