MongoDB Aggregation, group by subobject keys - mongodb

Question

I have a mongo collection whose schema looks like this:
_id: ObjectId(),
segments: {
activity: 'value1',
activation: 'value2',
plan: 'value3'
}
I'm trying to use the aggregation framework to find out how many of my documents have the value1 for the segment activity for instance.
The problem is that I want to do that for every segment in the same request if possible, and that I don't know how many segments I'll have or even their name.
Basically here's what I'd like to do:
If I have these two documents:
{ _id: 1, segments: { activity: 'active', activation: 'inactive', plan: 'free' }
{ _id: 2, segments: { activity: 'inactive', activation: 'inactive', plan: 'free' }
I want to be able to see that two of them have the activation segment to inactive and the free plan, and that activity have 1 inactive and 1 active values. Here is what I want to get:
{
activity: {
active: 1,
inactive: 1
},
activation: {
inactive: 2
},
plan: {
free: 2
}
}
So basically, if you could just $group by key it would be great! Something like this:
{
$group: {
_id: { $concat: [ '$segments.$key', '-', '$segments.$key.$value' ],
count: { $sum: 1 }
}
}
Or if I could unwind on each key...

Answer 1

To get the counts, take advantage of the $cond operator in the $group pipeline step to evaluate the counts based on the subdocuments value, something like the following:
db.collection.aggregate([
{
"$group": {
"_id": "$_id",
"activity_active": {
"$sum": {
"$cond": [ { "$eq": [ "$segment.activity", "active" ] }, 1, 0 ]
}
},
"activity_inactive": {
"$sum": {
"$cond": [ { "$eq": [ "$segment.activity", "inactive" ] }, 1, 0 ]
}
},
"activation_active": {
"$sum": {
"$cond": [ { "$eq": [ "$segment.activation", "active" ] }, 1, 0 ]
}
},
"activation_inactive": {
"$sum": {
"$cond": [ { "$eq": [ "$segment.activity", "inactive" ] }, 1, 0 ]
}
},
"plan_free": {
"$sum": {
"$cond": [ { "$eq": [ "$segment.plan", "free" ] }, 1, 0 ]
}
}
}
},
{
"$project": {
"_id": 0,
"activity": {
"active": "$activity_active",
"inactive": "$activity_inactive"
},
"activation": {
"active": "$activation_active",
"inactive": "$activation_inactive"
},
"plan": {
"free": "$plan_free"
}
}
}
])

Answer 2

there could be a generic solution to this problem, but might need a bit post processing:
to get output similat to this:
{
"_id" : {
"activity" : "active",
"activation" : "inactive"
},
"plan" : [{
"type" : "free",
"total" : 1
}, {
"type" : "paid",
"total" : 1
}
]
}, {
"_id" : {
"activity" : "inactive",
"activation" : "inactive"
},
"plan" : [{
"type" : "free",
"total" : 1
}
]
}, {
"_id" : {
"activity" : "inactive",
"activation" : "active"
},
"plan" : [{
"type" : "paid",
"total" : 3
}, {
"type" : "free",
"total" : 6
}
]
}
use query like that:
db.collection.aggregate([{
$group : {
_id : {
activity : "$segments.activity",
activation : "$segments.activation",
plan : "$segments.plan"
},
total : {
$sum : 1
}
}
}, {
$group : {
_id : {
activity : "$_id.activity",
activation : "$_id.activation"
},
plan : {
$push : {
type : "$_id.plan",
total : "$total"
}
}
}
},
])