MongoDB Aggregate subfields with individual date ranges - database

Question

I have a schema which is:
{
"_id" : "12345678",
"action1" : [
{
"date" : "2021-01-15",
"value" : 20
},
{
"date" : "2021-01-14",
"value" : 16
}
],
"action2" : [
{
"date" : "2021-01-15",
"value" : 30
},
{
"date" : "2021-01-14",
"value" : 10
}
],
"action3" : [
{
"date" : "2021-01-15",
"value" : 40
},
{
"date" : "2021-01-14",
"value" : 20
}
],
"action4" : [
{
"date" : "2021-01-15",
"value" : 60
},
{
"date" : "2021-01-14",
"value" : 40
}
]
}
Now I want to write an aggregate query to filter out counts within a date range (for last 7 days, or 30 days or 90 days)
so the final sum should look something like the following:
{
_id: "12345678"
action1 : {
alltime: number,
last7Days : number,
last30Days: number,
last90Days: number
},
action2: {
alltime: number,
last7Days : number,
last30Days: number,
last90Days: number
},
action3: {
alltime: number,
last7Days : number,
last30Days: number,
last90Days: number
},
action4: {
alltime: number,
last7Days : number,
last30Days: number,
last90Days: number
},
}
I am trying to get the total number of actions using $project and $match for a particular _id
But how can I filter the past7days/30days/90days data
My query looks like the following
db.collection.aggregate([
{
$project: {
alltimeAction1: {$sum: "$action1.value"},
alltimeAction2: {$sum: "$action2.value"},
alltimeAction3: {$sum: "$action3.value"},
alltimeAction4: {$sum: "$action4.value"}
}
},
{
$match: {
_id: "12345678"
}
}
])
is mapReduce the only option available?

Answer 1

db.test1.aggregate([
{
"$project": {//Reshape actions, you need this as you have dynamic keys
data: {
"$objectToArray": "$$ROOT"
}
}
},
{//Denormalize
"$unwind": "$data"
},
{//Denormalize
"$unwind": "$data.v"
},
{
"$project": {//Formatting date
"_id": 1,
"key": "$data.k",
"date": {
"$dateFromString": {
"dateString": "$data.v.date",
"format":"%Y-%m-%d"
}
},
"value": "$data.v.value"
}
},
{
"$addFields": {//Getting the conditions ready
"7Days": {
$gte:["$date", new Date(new Date().getTime() - (7*24*3600*1000))]
},
"30Days": {
$gte:["$date", new Date(new Date().getTime() - (30*24*3600*1000))]
}
}
},
{
$group:{//Grouping them, you can add few more cases
"_id": "$key",
"7days":{
$sum:"$value"
},
"30days":{
$sum:"$value"
}
}
}
])
You don't need to do this much complex query if you have flattened schema where actionName can be identified by a constant field name.