Get documents within a distance using Geospatial Query without Point object - mongodb

Question

I have a places collection that store location plainly as
place = {
name : "",
latitude: "",
longitude:""
}
Is there any way using mongo shell or spring data mongo where I can query places like this :
select all places with coordinates(places.longitude, place.latitude) near a point(x,y) and within a distance z . Something like:
db.places.find( {
{
"type" : "Point",
"coordinates" : [
places.longitude,
places.latitude
]
}:
{ $geoWithin:
{ $centerSphere: [ [ x, y ] ,z / 3963.2 ]
}
}
})
Or will I have to modify my collection to
place = {
name : "",
"loc" : {
"type" : "Point",
"coordinates" : [
longitude,
latitude
]
}
}

Answer 1

You really should change your data. MongoDB supports either a legacy coordinate pairs format or GeoJSON for geopatial indexes and queries only. You cannot use different fields for the data or "transform" in any way, as the supported field format is required by the "index" that is necessary for operations using $near or $nearSphere.
Best to to the transformation in the shell, since writing other API code for a "one off" operation is unnecesesary. And yes moving forward you really should be using the GeoJSON format:
var bulk = db.places.initializeUnorderedBulkOp(),
count = 0;
db.places.find().forEach(function(doc) {
bulk.find({ "_id": doc._id }).updateOne({
"$set": {
"location": {
"type": "Point",
"coordinates": [parseFloat(doc.longitude),parseFloat(doc.latitude)]
}
},
"$unset": { "latitude": "", "longitude": "" }
});
count++;
if ( count % 1000 == 0 ) {
bulk.execute();
bulk = db.places.initializeUnorderedBulkOp();
}
});
if ( count % 1000 !=0 )
bulk.execute();
Now the data is fixed and compatible with an index, create the index. What makes sense here with GeoJSON data is a "2sphere" index:
db.places.createIndex({ "location": "2dsphere" })
After that then you can query on the document as normal:
db.places.find({
"location": {
"$geoWithin": {
"$centerSphere": [ [ x, y ] ,z]
}
}
})
I should also note that a $centreSphere operation in a $geoWithin actually works out to be the same operation as $nearSphere with the $maxDistance modifier. The exception being that the latter should both process "faster" as well as produce "ordered" results for the "nearest" locations, which is something $geoWithin does not do:
db.places.find({
"$nearSphere": {
"$geometry": {
"type": "Point",
"coordinates": [x,y]
},
"$maxDistance": z
}
})
The only way you can do this on your existing data is for a $geoWithin only. This is because that operation does not require an geospatial index, so you are alowed to "transform" the document first.
You can do this using the .aggregate() method and it's $project pipeline stage along with the $map operator:
db.places.aggregate([
{ "$project": {
"name": 1
"location": {
"type": "Point",
"coordinates": {
"$map": {
"input": ["A","B"],
"as": "el",
"in": {
"$cond": [
{ "$eq": [ "$$el", "A" ] },
"$longitude",
"$latitude"
]
}
}
}
}
}},
{ "$match": {
"location": {
"$geoWithin": {
"$centerSphere": [ [ x, y ] ,z]
}
}
}}
])
However your longitude and latitude data must be numeric already as this is something you cannot transform in the aggregation framework. And you must remember that this cannot be used for operations such as $nearSphere as the required index is not available after the initial pipeline stage.
So it can be done, but it is not advisable. It's going to add processing time, and things are going to be better, more flexible and "faster" if you fix the data and add the appropriate index instead.
Also note that all distances with GeoJSON data will be in kilometers rather than radians.