MongoDB Query aggreate/group by and geometry/location/geoNear query - mongodb

Question

I am trying to fetch the documents with geometry within a certain location, however only want to return a single document per UUID. For this project, in most cases, there are many documents for each UUID that match the $near selector, hence we get many documents with the same UUID.
Can anyone assist with completing the below query so it only returns a single document per uuid (most recent "date")?
db.device.find(
{
location:
{ $near :
{
$geometry: { type: "Point", coordinates: [ -73.9667, 40.78 ] },
$minDistance: 1000,
}
}
}
)
Here's an example of the collection:
{
"_id":ObjectId("5a4f1ff0fc6ded723265e6b0"),
"uuid":"user1",
"date": "2018-01-20 11:58:29.000",
"location":{
"type": "Point",
"coordinates":[
//remove for demo sake
]
}
},
{
"_id":ObjectId("5a62a245ce689f68245450a7"),
"uuid":"user2",
"date": "2018-01-20 11:58:07.000",
"location":{
"type": "Point",
"coordinates":[
//remove for demo sake
]
}
},
{
"_id":ObjectId("5a62a20fce689f7a14648c62"),
"uuid":"user1",
"date": "2018-01-20 11:58:39.000",
"location":{
"type": "Point",
"coordinates":[
//remove for demo sake
]
}
},
{
"_id":ObjectId("5a62a205ce689f7039203923"),
"uuid":"user1",
"date": "2018-01-20 11:58:49.000",
"location":{
"type": "Point",
"coordinates":[
//remove for demo sake
]
}
},
{
"_id":ObjectId("5a62a277ce689f703a3eacb3"),
"uuid":"user2",
"date": "2018-01-20 11:58:59.000",
"location":{
"type": "Point",
"coordinates":[
//remove for demo sake
]
}
}

Answer 1

When performing this kind of heavier operations, you can switch to using an aggregation pipeline.
Using this input:
{
"uuid": "user1",
"date": "2018-01-20 11:58:29.000",
"location": { "type": "Point", "coordinates":[-0.17818, 51.15609] }
},
{
"uuid": "user2",
"date": "2018-01-20 11:58:07.000",
"location": { "type": "Point", "coordinates":[2.35944, 48.72528] }
},
{
"uuid": "user1",
"date": "2018-01-20 11:58:39.000",
"location": { "type": "Point", "coordinates": [1.45414, 43.61132] }
},
{
"uuid": "user1",
"date": "2018-01-20 11:58:49.000",
"location": { "type": "Point", "coordinates":[-6.24889, 53.33306] }
},
{
"uuid": "user2",
"date": "2018-01-20 11:58:59.000",
"location": { "type": "Point", "coordinates":[-3.68248, 40.47184] }
}
Using this index:
db.device.createIndex( { location : "2dsphere" } )
This pipeline should perform what you want:
db.device.aggregate([
{ $match: { location: { $geoWithin: { $centerSphere: [ [ -0.17818, 51.15609 ], 0.1232135647961246 ] } } } },
{ $sort: { "date": -1 } },
{ $group: { _id: { uuid: "$uuid" }, users: { $push: { "uuid": "$uuid", "date": "$date", "location": "$location" } } } },
{ $project: { user: { $arrayElemAt: ["$users", 0] } } }
])
I first adapted the find/$near operator to an aggregation equivalent ($geoWithin/$centerSphere). It matches locations within 0.123 radians (488 kms (0.123*3963.2)).
I then directly sort by date, this way when documents will then be grouped by user, I will be able to easily select the first per user.
I then group by user.
And finally for each user, as I have a value produced by $group which is an array of the user documents (sorted), I just extract the first item of the array with $arrayElemAt.
This produces:
{
"_id" : { "uuid" : "user2" },
"user": {
"uuid": "user2",
"date": "2018-01-20 11:58:07.000",
"location": { "type": "Point", "coordinates": [ 2.35944, 48.72528 ] }
}
}
{
"_id": { "uuid" : "user1" },
"user": {
"uuid": "user1",
"date": "2018-01-20 11:58:49.000",
"location": { "type": "Point", "coordinates": [ -6.24889, 53.33306 ] }
}
}