Mongodb aggregate: convert date to another timezon

Question

I save my transaction with something like :

{code: "A", total: 250000, timestamp: ISODate("2016-01-20T23:57:05.771Z")},
{code: "B", total: 300000, timestamp: ISODate("2016-01-20T05:57:05.771Z")}

each of transaction has timestamp field under UTC timezone in it. Since I live in Jakarta (UTC+7) timezone, I need to add 7 hours to my timestamp before aggregation. Here's my mongo syntax:

db.transaction.aggregate(
[
 {
   $project:
     {
       year: { $year: "$timestamp" },
       month: { $month: "$timestamp" },
       day: { $dayOfMonth: "$timestamp" }
     }
 }
])

It returns:

    {
        "_id" : ObjectId("56a01ed143f2fd071793d63b"),
        "year" : 2016,
        "month" : 1,
        "day" : 20
    },
    {
        "_id" : ObjectId("56a01ed143f2fd071793d63b"),
        "year" : 2016,
        "month" : 1,
        "day" : 20
    }

which is wrong since the first transaction (code A), is happened at 21st January, but since it was converted to UTC (-7 Hours), it became ISODate("2016-01-20T23:57:05.771Z")

Note: I'm aware about the same problem over here, here's what I've been tried so far:

db.transaction.aggregate(
[
 {$project: {"timestamp": {$add: [7 * 60 * 60 * 1000]}}},
 {
   $project:
     {
       year: { $year: "$timestamp" },
       month: { $month: "$timestamp" },
       day: { $dayOfMonth: "$timestamp" }
     }
 }
])

but it returns can't convert from BSON type NumberDouble to Date error. Any suggestions?

Answer 1

You need to add the "timestamp" to 7 * 60 * 60 * 1000 also you can do this in one $project stage.

db.collection.aggregate([
    { "$project": {
        "year": { "$year": { "$add": [ "$timestamp", 7 * 60 * 60 * 1000 ] } }, 
        "month": { "$month": { "$add": [ "$timestamp", 7 * 60 * 60 * 1000 ] } }, 
        "day": { "$dayOfMonth": { "$add": [ "$timestamp", 7 * 60 * 60 * 1000 ] } } 
    } }
])

Which returns:

{
        "_id" : ObjectId("56a07f8f8448ed9fd2359365"),
        "year" : 2016,
        "month" : 1,
        "day" : 21
}
{
        "_id" : ObjectId("56a07f8f8448ed9fd2359366"),
        "year" : 2016,
        "month" : 1,
        "day" : 20
}

Answer 2

As an update, MongoDB 3.6 has a new timezone parameter for date manipulation in the aggregation framework. Most date-related operators accept this optional parameter, see $hour for one example.

For example, if we have a document where the date is exactly the new year in UTC:

> db.test.find()
{"_id": 1, "dt": ISODate("2018-01-01T00:00:00Z")}

We can display the date in New York timezone:

> db.test.aggregate([
...     {$project:{
...         date:{$dayOfMonth:{date:'$dt',timezone:'America/New_York'}},
...         month:{$month:{date:'$dt',timezone:'America/New_York'}},
...         year:{$year:{date:'$dt',timezone:'America/New_York'}},
...         hour:{$hour:{date:'$dt',timezone:'America/New_York'}}
...     }}
... ])
{ "_id": 1, "date": 31, "month": 12, "year": 2017, "hour": 19 }

We can also display the date in Sydney timezone:

> db.test.aggregate([
...     {$project:{
...         date:{$dayOfMonth:{date:'$dt',timezone:'Australia/Sydney'}},
...         month:{$month:{date:'$dt',timezone:'Australia/Sydney'}},
...         year:{$year:{date:'$dt',timezone:'Australia/Sydney'}},
...         hour:{$hour:{date:'$dt',timezone:'Australia/Sydney'}}
...     }}
... ])
{ "_id": 1, "date": 1, "month": 1, "year": 2018, "hour": 11 }

The timezone description is using the standard Olson Timezone Identifier string.