MongoDB Aggregation by Shifted Date - mongodb

Question

Is there a way to aggregate by day, but over a 24 hour period that does not go from 12am - 11:59pm? A sample document looks like this:
{
date: ISODate("2012-11-02T17:04:11.102Z"),
user: 'testUser',
orders: 50
}
I need to aggregate the # of orders per user per day between 5 pm and 4:59:59.999 pm the next day. I can get the # of orders per user per day (over a 2 day range) using this:
db.hs.aggregate([{
$match: {
user: 'testUser',
date: {
$gte: new Date(2015,0,4,17,0,0,0),
$lt: new Date(2015,0,6,17,0,0,0)
}
}
}, {
$group: {
_id: {
date: {
month: {$month: "$date"},
day: {$dayOfMonth: "$date"},
year: {$year: "$date"}
},
user: "$user",
},
totord: {$sum: "$orders"}
}
}])
But this returns 3 results, one for Jan 4 after 5pm, one for Jan 5 all day, and one for Jan 6 before 5pm. I just don't know how to shift the aggregation to be between 5pm - 4:59pm.

Answer 1

I believe that you could use the $add aggregation operator on your dates in a $project aggregation step to get the desired result.
db.hs.aggregate([
{
$match: {
user: 'testUser',
date: {
$gte: new Date(2015,0,4,17,0,0,0),
$lt: new Date(2015,0,6,17,0,0,0)
}
}
}, {
$project: {
orders: 1,
user: 1,
date: { $add: [ "$date", 7*60*60000 ] }
}
}, {
$group: {
_id: {
date: {
month: { $month: "$date" },
day: { $dayOfMonth: "$date" },
year: { $year: "$date" }
},
user: "$user"
},
totord: {$sum: "$orders"}
}
}
])
I believe this should add 7 hours to the $date which should make it so anything after 5PM ends up after midnight the next day.