How to get single document with group that have duplicate key in MongoDB - mongodb

Question

I have an aggregate as follows:
[
{
"$project" : {
"country_code" : "$country_code",
"event" : "$event",
"user_id" : "$user_id",
"os" : "$os",
"register_time" : "$register_time",
"channel" : "$channel"
}
},
{
"$match" : {
"channel" : "000001",
"register_time" : {
"$gt" : ISODate("2016-06-01T00:00:00Z"),
"$lt" : ISODate("2016-06-30T23:59:00Z")
},
"event" : "Register_with_number"
}
},
{
"$group" : {
"_id" : {
"country_code" : "$country_code",
"user_id" : "$user_id",
"os" : "$os",
"channel" : "$channel",
"register_time" : "$register_time"
},
"count" : {
"$sum" : 1
}
}
}
]
And the result is as follows: you can for the country_code with IN, two records are having same user_id but different register_time, how can I get only one record if the user_id is same.
{ "_id" : { "country_code" : "US", "user_id" : "d2a0fe91", "os" : "Android", "channel" : "000001", "register_time" : ISODate("2016-06-30T22:47:43Z") }, "count" : 1 }
{ "_id" : { "country_code" : "US", "user_id" : "77911591", "os" : "Android", "channel" : "000001", "register_time" : ISODate("2016-06-30T19:47:21Z") }, "count" : 1 }
{ "_id" : { "country_code" : "IN", "user_id" : "1b72fd12", "os" : "Android", "channel" : "000001", "register_time" : ISODate("2016-06-30T19:17:28Z") }, "count" : 1 }
{ "_id" : { "country_code" : "IN", "user_id" : "1b72fd12", "os" : "Android", "channel" : "000001", "register_time" : ISODate("2016-06-30T19:15:13Z") }, "count" : 1 }
{ "_id" : { "country_code" : "ID", "user_id" : "045f1637", "os" : "Android", "channel" : "000001", "register_time" : ISODate("2016-06-30T19:02:19Z") }, "count" : 1 }

Answer 1

There are several solutions, as you did not mention what the document should look like when there are several documents with the same user but different register_time.
The following changes your last $group stage so that it keeps an array of the register_time values with $push or - if you just need one - keeps any of them with $first. Note that when you sort your pipeline by register_time, you could use $first / $last to keep the first / last register_time per user which is perhaps your desired result.
"$group" : {
"_id" : {
"country_code" : "$country_code",
"user_id" : "$user_id",
"os" : "$os",
"channel" : "$channel",
},
"register_times" : {
$push: "$register_time"
},
"any_register_time" : {
$first: "$register_time"
},
"count" : {
"$sum" : 1
}
}