How to handle unordered bucket boundaries in mongodb? - mongodb

Question

I need to retrieve all profile views a user received within 15 days. I made a working query that group all views by day.
A view is a very simple document:
{
viewerId: string,
username: string,
profileId: string,
day: 4
}
The day key matches the current day of the month when the view was stored. So I made this query:
static getProfileViews(userId: string) {
const fifteenDaysRange = this.getDaysRange(15);
return ProfileViewsDB.aggregate([
{ $match: { profileId: userId } },
{
$bucket: {
groupBy: "$day",
boundaries: fifteenDaysRange,
default: fifteenDaysRange[fifteenDaysRange.length - 1] + 1,
output: {
count: {
$sum: 1,
},
viewers: {
$push: {
_id: "$viewerId",
username: "$username",
},
},
},
},
},
]);
}
It works, except when the days range is across two months. So the boundaries are: [29 ,30, 31, 1, 2, 3, 4], instead of a regular incremental list. In this case, mongo crashes, claiming that the boundaries must be properly ordered.
How to fix this? Is there a trick here to handle this edge-case?

Answer 1

Not clear what you are looking for, but I assume it should work with this:
db.ProfileViewsDB.aggregate([
{ $match: { profileId: userId } },
{ $match: { date: { $gte: moment().startOf('day').subtract(15, 'day').toDate() } } },
{
$group: {
_id: { y: { $year: "$date" }, m: { $month: "$date" }, d: { $dayOfMonth: "$date" } },
count: { $sum: 1 },
viewers: {
$push: { _id: "$viewerId", username: "$username", },
}
}
}
])
Whenever one has to work with date/times then I recommend the moment.js library, it makes your life easier.
Note, you may add the timezone, i.e.
_id: {
y: { $year: "$date", timezone: "Europe/Zurich" },
m: { $month: "$date", timezone: "Europe/Zurich" },
d: { $dayOfMonth: "$date", timezone: "Europe/Zurich" }
}