Mongodb aggregate by day and delete duplicate value - mongodb

Question

I'm trying to clean a huge database.
Sample DB :
{
"_id" : ObjectId("59fc5249d5ab401d99f3de7f"),
"addedAt" : ISODate("2017-11-03T11:26:01.744Z"),
"__v" : 0,
"check" : 17602,
"lastCheck" : ISODate("2018-04-05T11:47:00.609Z"),
"tracking" : [
{
"timeCheck" : ISODate("2017-11-06T13:17:20.861Z"),
"_id" : ObjectId("5a0060e00f3c330012bafe39"),
"rank" : 2395,
},
{
"timeCheck" : ISODate("2017-11-06T13:22:31.254Z"),
"_id" : ObjectId("5a0062170f3c330012bafe77"),
"rank" : 2395,
},
{
"timeCheck" : ISODate("2017-11-06T13:27:40.551Z"),
"_id" : ObjectId("5a00634c0f3c330012bafebe"),
"rank" : 2379,
},
{
"timeCheck" : ISODate("2017-11-06T13:32:41.084Z"),
"_id" : ObjectId("5a0064790f3c330012baff03"),
"rank" : 2395,
},
{
"timeCheck" : ISODate("2017-11-06T13:37:51.012Z"),
"_id" : ObjectId("5a0065af0f3c330012baff32"),
"rank" : 2379,
},
{
"timeCheck" : ISODate("2017-11-07T13:37:51.012Z"),
"_id" : ObjectId("5a0065af0f3c330012baff34"),
"rank" : 2379,
}]
}
I have a lot of duplicate value but I need to clean only by day.
To obtain this for example :
{
"_id" : ObjectId("59fc5249d5ab401d99f3de7f"),
"addedAt" : ISODate("2017-11-03T11:26:01.744Z"),
"__v" : 0,
"check" : 17602,
"lastCheck" : ISODate("2018-04-05T11:47:00.609Z"),
"tracking" : [
{
"timeCheck" : ISODate("2017-11-06T13:17:20.861Z"),
"_id" : ObjectId("5a0060e00f3c330012bafe39"),
"rank" : 2395,
},
{
"timeCheck" : ISODate("2017-11-06T13:27:40.551Z"),
"_id" : ObjectId("5a00634c0f3c330012bafebe"),
"rank" : 2379,
},
{
"timeCheck" : ISODate("2017-11-07T13:37:51.012Z"),
"_id" : ObjectId("5a0065af0f3c330012baff34"),
"rank" : 2379,
}]
}
How can I aggregate by day and after delete last value duplicate?
I need to keep the values per day even if they are identical with another day.

Answer 1

The aggregation framework cannot update data at this stage. However, you can use the following aggregation pipeline in order to get the desired output and then use e.g. a bulk replace to update all your documents:
db.collection.aggregate({
$unwind: "$tracking" // flatten the "tracking" array into separate documents
}, {
$sort: {
"tracking.timeCheck": 1 // sort by timeCheck to allow us to use the $first operator in the next stage reliably
}
}, {
$group: {
_id: { // group by
"_id": "$_id", // "_id" and
"rank": "$tracking.rank", // "rank" and
"date": { // the "date" part of the "timeCheck" field
$dateFromParts : {
year: { $year: "$tracking.timeCheck" },
month: { $month: "$tracking.timeCheck" },
day: { $dayOfWeek: "$tracking.timeCheck" }
}
}
},
"doc": { $first: "$$ROOT" } // only keep the first document per group
}
}, {
$sort: {
"doc.tracking.timeCheck": 1 // restore ascending sort order - may or may not be needed...
}
}, {
$group: {
_id: "$_id._id", // merge everything again per "_id"
"addedAt": { $first: "$doc.addedAt" },
"__v": { $first: "$doc.__v" },
"check": { $first: "$doc.check" },
"lastCheck": { $first: "$doc.lastCheck" },
"tracking": { $push: "$doc.tracking" } // in order to join the tracking values into an array again
}
})