How to find events that occurred in a timeframe (mongo) - mongodb

Question

I have the following document structure:
{ _id:ID1
value: { data:{userData:{name:aaa,surname:bbb}}
events:[
{even1tName:{timestamp:UNIX_TIMESTAMP,value:NUMBER}},
{even2tName:{timestamp:UNIX_TIMESTAMP,value:NUMBER}},
{even3tName:{timestamp:UNIX_TIMESTAMP,value:NUMBER}},
{even4tName:{timestamp:UNIX_TIMESTAMP,value:NUMBER}},
],
activity:{countEvents:INTEGER,totalValue:NUMBER}
}
}
This is the output of a MapReduce pipe, I need to find using aggregation, what users have a certain amount of events and a certain amount of value (summed up), within a timeframe. Consider these are online buyers and I need to find those that have made 3 purchases within the last month or those that have bought of a total amount greater than $300.

Answer 1

Your question is a bit light on information, but the main thing is that as long as there is consistent "keyname" naming in the documents then this really is not an issue:
db.junk.aggregate([
// Match where type within timeframe
{ "$match": {
"value.events.confirmedSale.timestamp": {
"$gte": startTime, "$lt": endTime
}
}},
// Pre-filter the array for required data
{ "$project": {
"value": {
"data": "$value.data",
"events": {
"$setDifference": [
{"$map": {
"input": "$value.events",
"as": "el",
"in": {
"$cond": [
{ "$and": [
{ "$gte": [ "$$el.confirmedSale.timestamp", startTime ] },
{ "$lt": [ "$$el.confirmedSale.timestamp", endTime ] }
]},
"$$el",
false
]
}
}},
[false]
]
}
}
}},
// Unwind array elements for processing
{ "$unwind": "$value.events" },
// Group data
{ "$group": {
"_id": "$_id",
"value": { "$sum": "$value.events.confirmedSale.value"},
"count": { "$sum": 1 }
}},
// Filter results on totals
{ "$match": {
"value": { "$gte": 300, "count": { "$gte": 3 } }
}}
])
However, due to the document structure you cannot really get more extensive than that. Such naming requires "path names" to embedded objects to be absolute, and this particular case does not do well for indexing either.
With some control over the document creation, then it should look more like this:
{ _id: 1,
value: {
data:{
userData:{name:"aaa",surname:"bbb"}
},
events:[
{ "type": "adCLick", "timestamp": 1234, "value": 1234 },
{ "type": "confirmedSale", "timestamp": 5678, "value": 5678 },
{ "type": "confirmedSale", "timestamp": 4567, "value": 4567 },
{ "type": "something", "timestamp": 9876, "value": 9876}
]
}
}
Now that the field Name you were using here is actually now just a consistent "data" property, the query can be more clearly readable, do more with combined events that you cannot do, and also work in the use of indexes for performance.
MongoDB is primarily a "database", if you do not keep consistent naming paths then you will have performance and feature loss as a consequence. The aggregation framework is the "high performance" option over mapReduce with JavaScript. Working with a set key pattern is fine for the aggregation framework, but if you vary that pattern, then your only option is mapReduce.