MongoDB: Aggregation Group operation over a dynamic key-value pair - mongodb

Question

I have a simple document with one field as a key-value pair. I want to just perform a group operation in Aggregation over those keys and add their values. But the keys in the pair are not fixed and can be anything.
Here is a sample document.
{
_id: 349587843,
matchPair: {
3 : 21,
9 : 4,
7 : 32
}
},
{
_id: 349587478,
matchPair: {
7 : 11,
54 : 32,
9 : 7,
2 : 19
}
}
And I want a result something like the following.
{
_id : 3,
count : 21
},
{
_id : 9,
count : 11
},
{
_id : 7,
count : 43
},
{
_id : 54,
count : 32
},
{
_id : 2,
count : 19
}
I have the following query in mind and tried using $unwindoperation but it doesn't work probably because "matchPair" isn't an array and I don't know what to specify for the $sumoperation.
db.MatchPairs.aggregate([
{ "$unwind" : "$matchPair" },
{ "$group" : {
_id: "$matchPair",
count : { $sum : $matchPair }
} }
]);
I could also try Map-Reduce but for that too I need to emit() keys and values by name.
I'm sure there's a simple solution to this but I can't figure it out.

Answer 1

:
You could start by projecting and reshaping the matchPair field with $objectToArray
New in version 3.4.4.
{
$project: {
matchPair: { $objectToArray: '$matchPair' }
}
}
which would give
{
matchPair: [{ k: 3, v: 21 }, { k: 9, v: 4 }, ...]
}
Then $unwind based on matchPair
{
$unwind: '$matchPair'
}
which would give
{
matchPair: { k: 3, v: 21 }
}
Then $project
{
$project: {
_id: '$matchPair.k',
count: '$matchPair.v'
}
}
That should give the output you want. Altogether would be
.aggregate([
{
$project: {
matchPair: { $objectToArray: '$matchPair' }
}
},
{ $unwind: '$matchPair' },
{
$project: {
_id: '$matchPair.k',
count: '$matchPair.v'
}
}
])

Answer 2

In the mongoDb documentation for $unwind:
Deconstructs an array field from the input documents to output a
document for each element.
So you have to change your schema for something like:
{
"_id" : ObjectId("5880b57b039a3c89c1db145a"),
"matchPair" : [
{
"_id" : "3",
"count" : 21
},
{
"_id" : "9",
"count" : 4
},
{
"_id" : "7",
"count" : 32
}
]
},
{
"_id" : ObjectId("5880b58c039a3c89c1db145b"),
"matchPair" : [
{
"_id" : "7",
"count" : 11
},
{
"_id" : "54",
"count" : 32
},
{
"_id" : "9",
"count" : 7
},
{
"_id" : "2",
"count" : 19
}
]
}
Then doing:
db.MatchPairs.aggregate([
{ $unwind : "$matchPair" }
]);
will return:
{
"_id" : ObjectId("5880b57b039a3c89c1db145a"),
"matchPair" : {
"_id" : "3",
"count" : 21
}
},
{
"_id" : ObjectId("5880b57b039a3c89c1db145a"),
"matchPair" : {
"_id" : "9",
"count" : 4
}
},
{
"_id" : ObjectId("5880b57b039a3c89c1db145a"),
"matchPair" : {
"_id" : "7",
"count" : 32
}
},
{
"_id" : ObjectId("5880b58c039a3c89c1db145b"),
"matchPair" : {
"_id" : "7",
"count" : 11
}
},
{
"_id" : ObjectId("5880b58c039a3c89c1db145b"),
"matchPair" : {
"_id" : "54",
"count" : 32
}
},
{
"_id" : ObjectId("5880b58c039a3c89c1db145b"),
"matchPair" : {
"_id" : "9",
"count" : 7
}
},
{
"_id" : ObjectId("5880b58c039a3c89c1db145b"),
"matchPair" : {
"_id" : "2",
"count" : 19
}
}
Then you just have to do your grouping.