How to project additional data from an aggregate result with MongoDB? - mongodb

Question

I'm learning MongoDB and try to group a collection.
What I'm looking for is to group by year, get the max "average note" field and display the field primary name of the document related to this average
For example, if I have:
Name | Average | Year
Name_01 | 7.56 | 1995
Name_02 | 8.96 | 1995
Name_03 | 3.25 | 2005
Name_04 | 4.36 | 2005
Name_05 | 7.52 | 2020
I need:
Name | Average | Year
Name_02 | 8.96 | 1995
Name_05 | 7.52 | 2020
Name_04 | 4.36 | 2005
I already did the group and the max. Here is my code:
db.foobar.aggregate([
{
$group: { _id: '$year_published', max: { $max: '$statistics.average' }}
},
{
$project: { _id: 1, max: 1 }
},
{
$sort: { max: -1 }
}
])
Which gives me this kind of result:
{
"result" : [
{
"_id" : 1999,
"max" : 8.0343000000000000
},
{
"_id" : 1985,
"max" : 7.8833299999999999
}
// An so on...
}
But I'd also like to project the primary name of the document related to the "max" to get something like:
{
"result" : [
{
"_id" : 1999,
"max" : 8.0343000000000000,
"name": "Foo Bar"
},
{
"_id" : 1985,
"max" : 7.8833299999999999,
"name": "Lorem Ipsum"
}
// An so on...
}
NB : The next part of the question add complexity regarding the name (because of my document structure). It's not my main concern right now, but I add it to the question to reflect all my problem.
The primary name is a bit tricky to get. For each document, I've got an array of objects like that:
{
"names" : [
{
"type" : "primary",
"value" : "Foo bar"
},
{
"type" : "alternate",
"value" : "Foo foo"
},
{
"type" : "alternate",
"value" : "Bar bar"
}
]
}
And what I'm trying to get is the name with "primary" type (i. e. "Foo bar" in my example).
Here is the structure of my documents:
{
"_id" : ObjectId("56338f2bdc99b8ec22a43328"),
"names" : [
{
"type" : "primary",
"value" : "Foo bar"
},
{
"type" : "alternate",
"value" : "Barr foo"
}
],
"year_published" : 1992
"statistics" : {
"average" : 6.6057699999999997
}
}
I think I'm not so far but I don't know how to do it... Could you please help me?

Answer 1

If you want the "paried" values out of a particular doccument with a "max" value then $max is not for you. Instead what you need to do is $sort the data first and then use the $first operator.
db.foobar.aggregate([
{ "$sort": { "year_published": 1, "statistics.average": -1 } },
{ "$group": {
"_id": "$year_published",
"max": { "$first": "$statistics.average" }},
"name": {
"$first": {
"$setDifference": [
{ "$map": {
"input": "$names",
"as": "name",
"in": {
"$cond": {
"if": { "$eq": [ "$$name.type", "primary" ] },
"then": "$$name.value",
"else": false
}
}
}},
[false]
]
}
}
}},
{ "$unwind": "$name" }
])
The $first and $last operators act on "grouping boundary" data. Which means they return data from the property that occurs either at the begining or end of the value that was used for the grouping _id.
That is why you "sort" first, so th documents are in order for selection.
By contrast $max and $min just pick the "max/min" value from anywhere in the documents in the sample. That's fine when it's all you want, but if you want "related" fields, then you must sort first.
That's the basics of it. The other part for dealing with filtering your array is most optimally done with the $map and $setDifference combination as shown. The $map allows testing of a condition via $cond on each array element "in-line", and returns the value depending on which is true or false. The result is still of course an array of equal length.
The $setDifference essentially filters out anything returned as false, so the only thing left should be the "primary". Still an array, which is why $unwind is still used, though it's only a single element array.
Future MongoDB versions will do this a little better with $filter and $arrayElemAt. Here's a glimpse:
db.foobar.aggregate([
{ "$sort": { "year_published": 1, "statistics.average": -1 } },
{ "$group": {
"_id": "$year_published",
"max": { "$first": "$statistics.average" }},
"name": {
"$first": {
"$arrayElemAt": [
{ "$filter": {
"input": "$names",
"as": "name",
"cond": {
"$eq": [ "$$name.type", "primary" ]
}
}},
0
]
}
}
}}
])
But none of this changes the basic rules of "sort first" and then just pick up the values from the grouping boundary.

Answer 2

Please try the below code :
You need to select the "name" filed in the group pipeline operation with the help of $First.
$First selects the value that results from applying an expression to the first document in a group of documents that share the same group by key.
db.foobar.aggregate([
{ "$unwind" : "$names" },
{ $match :
{ "$names.type" : "primary"}
} ,
{ $sort :
{ "year_published" : 1, "statistics.average" : -1 }
},
{ $group :
{ _id : "$year_published" ,
name : {
$first : "$names.value"
},
max: { $max: "$statistics.average" }
}
},
{ $sort:
{ max: -1 }
}
]).pretty();
This will give you the required result :
{
"result" : [
{
"_id" : 1999,
"max" : 8.0343000000000000,
"name": "Foo Bar"
},
{
"_id" : 1985,
"max" : 7.8833299999999999,
"name": "Lorem Ipsum"
}
// An so on...
}