Group across two columns in Mongo - mongodb

Question

Say in mongo I have a collection that looks like this:
+----+-----+-----+----------+
| id | x | y | quantity |
+----+-----+-----+----------+
| 1 | abc | jkl | 5 |
+----+-----+-----+----------+
| 2 | jkl | xyz | 10 |
+----+-----+-----+----------+
| 3 | xyz | abc | 20 |
+----+-----+-----+----------+
I want to do a $group where x equals y and sum up the quantity. So the output would look like:
+-----+-------+
| x | total |
+-----+-------+
| abc | 25 |
+-----+-------+
| jkl | 15 |
+-----+-------+
| xyz | 30 |
+-----+-------+
Is this even possible to do in mongo?

Answer 1

You won't be performing a $group to retrieve the results. You're performing a $lookup. This feature is new in MongoDB 3.2.
Using the sample data you provided, the aggregation would be the following:
db.join.aggregate( [
{
"$lookup" : {
"from" : "join",
"localField" : "x",
"foreignField" : "y",
"as" : "matching_field"
}
},
{
"$unwind" : "$matching_field"
},
{
"$project" : {
"_id" : 0,
"x" : 1,
"total" : { "$sum" : [ "$quantity", "$matching_field.quantity"]}
}
}
])
The sample data set is pretty simple, so you'll need to test behavior when there are more than a simple result returned for a value, etc.
Edit:
It gets more complicated if there can be more than a single match between X and Y.
// Add document to return more than a single match for abc
db.join.insert( { "x" : "123", "y" : "abc", "quantity" : 100 })
// Had to add $group stage to consolidate matched results
db.join.aggregate( [
{
"$lookup" : {
"from" : "join",
"localField" : "x",
"foreignField" : "y",
"as" : "matching_field"
}
},
{
"$unwind" : "$matching_field"
},
{ "$group" : {
"_id" : { "x" : "$x", "quantity" : "$quantity" },
"matched_quantities" : { "$sum" : "$matching_field.quantity" }
}},
{
"$project" : {
"x" : "$_id.x",
"total" : { "$sum" : [ "$_id.quantity", "$matched_quantities" ]}
}
}
])