MongoDB, grouping inner Hash key value by another key value - mongodb

Question

I have these 4 elements in my collection:
/* 1 */
{
"demographics": [
{
"key": "country",
"value": "ES"
},
{
"key": "city",
"value": "Sevilla"
},
{
"key": "region",
"value": "Andalucía"
}
]
}
/* 2 */
{
"demographics": [
{
"key": "city",
"value": "Cádiz"
},
{
"key": "country",
"value": "ES"
},
{
"key": "region",
"value": "Andalucía"
}
]
}
/* 3 */
{
"demographics": [
{
"key": "country",
"value": "GB"
},
{
"key": "region",
"value": "Greater London"
},
{
"key": "city",
"value": "London"
}
]
}
/* 4 */
{
"demographics": [
{
"key": "country",
"value": "ES"
},
{
"key": "region",
"value": "Andalucía"
},
{
"key": "city",
"value": "Sevilla"
}
]
}
I would like to group them by:
demographic.value when demographic.key = "country"
demographic.value when demographic.key = "region"
demographic.value when demographic.key = "city"
Having a result like this:
{ "values": ["ES", "Andalucía", "Sevilla"], "count": 2 }
{ "values": ["ES", "Andalucía", "Cádiz"], "count": 1 }
{ "values": ["GB", "Greater London", "London"], "count": 1 }
Attention: beware the order of the demographics array elements might be not always the same.
I have tried
db.getCollection('test').aggregate(
[
{ "$unwind": "$demographics" },
{
"$project" :{
"_id": 0,
"demographics.key": 1,
"demographics.value": 1
}
},
{
"$group" : {
"_id": {
"key": "$demographics.key",
"value": "$demographics.value"
},
"count": { "$sum": 1 }
}
},
{
"$group" : {
"_id": "$_id.key",
"values": { "$push": { "value": "$_id.value", "count": "$count" } }
}
}
]
)
This gives me this result:
/* 1 */
{
"_id": "country",
"values": [
{
"value": "GB",
"count": 1.0
},
{
"value": "ES",
"count": 3.0
}
]
}
/* 2 */
{
"_id": "region",
"values": [
{
"value": "Greater London",
"count": 1.0
},
{
"value": "Andalucía",
"count": 3.0
}
]
}
/* 3 */
{
"_id": "city",
"values": [
{
"value": "London",
"count": 1.0
},
{
"value": "Cádiz",
"count": 1.0
},
{
"value": "Sevilla",
"count": 2.0
}
]
}
But this is not the groups I am looking for

Answer 1

You can try running the following pipeline:
db.test.aggregate([
{ "$unwind": "$demographics" },
{ "$sort": { "demographics.key": 1, "demographics.value": 1 } },
{
"$group": {
"_id": "$_id",
"values": { "$push": "$demographics.value" }
}
},
{
"$group": {
"_id": "$values",
"count": { "$sum": 1 }
}
},
{
"$project": {
"_id": 0, "values": "$_id", "count": 1
}
}
])
Sample Output
/* 1 */
{
"count" : 2,
"values" : [
"Sevilla",
"ES",
"Andalucía"
]
}
/* 2 */
{
"count" : 1,
"values" : [
"London",
"GB",
"Greater London"
]
}
/* 3 */
{
"count" : 1,
"values" : [
"Cádiz",
"ES",
"Andalucía"
]
}