How do I use a geospatial query in the 2.1 MongoDB

2020-08-09 18:53发布

问题:

I've been banging my head on this one for days. I have a very simple query I'm trying to run in C#, it looks like this in the shell.

db.runCommand({geoNear: "items", near: {type: "Point", coordinates : [-111.283344899999, 47.4941836]}, spherical : true, distanceMultiplier: 3963.2, maxDistance : 25});

My collection looks like this

{    
  "_id" : ObjectId(),    
  "Title" : "arst",    
  "Description" : "<p>arst</p>",    
  "Date" : new Date("11/29/2015 09:28:15"),    
  "Location" : {    
    "type" : "Point",    
    "Coordinates" : [-111.28334489999998, 47.4941836]    
  },    
  "Zip" : "59405"    
}

According to the docs here MongoDB C# API Docs the MongoDB.Driver.Builders.Query object is now legacy. So when I do something like this

 var point = new GeoJson2DGeographicCoordinates(double.Parse(longitude), double.Parse(latitude)) ;
        var query = Query<Item>.Near(x => x.Location, new GeoJsonPoint<GeoJson2DGeographicCoordinates>(point), distance, true);

        var result = collection.Find(query);

The compiler complains that it can't convert from IMongoQuery to FilterDefinition. This tells me that the legacy Query<> builder isn't supported by the new 2.1 library. But for the life of me I can't find anywhere in the api docs that reference a replacement?

Can anyone point me in the right direction on executing this simple geo-spatial query in the 2.1 C# Driver? I'm stumped.

Also, I do have a 2dsphere index created on the collection, if I didn't the shell command wouldn't work. Here's the index output.

{
            "v" : 1,
            "key" : {
                    "Location.Coordinates" : "2dsphere"
            },
            "name" : "Location.Coordinates_2dsphere",
            "ns" : "ppn.items",
            "2dsphereIndexVersion" : 2
    }

EDIT

After digging through a TON of documentation I think I found it. All the examples still show the legacy Query<> method, but it seems that the new method is part of the Builders<>.Filter namespace. So this code block seems to be working for me,

 double lng = double.Parse(longitude);
 double lat = double.Parse(latitude);
 point = new GeoJson2DGeographicCoordinates(lng, lat);
 pnt = new GeoJsonPoint<GeoJson2DGeographicCoordinates>(point);
 dis = distance * 1609.34;
 fil = Builders<Item>.Filter.NearSphere(p => p.Location.Coordinates, pnt, dis);

 filter = filter & fil;

 var sort = Builders<Item>.Sort.Descending("Date");

 // This is the actual query execution
 List<Item> items = collection.Find(filter).Sort(sort).ToListAsync().Result;

This code block is very messy, it's the result of try and fail over and over. I'm sure I'll figure out ways to clean it up. It seems a little verbose to me that you have to create a GeoJsonPoint from a GeoJson2DGeographicCoordinates, but I'm sure there's a good reason for it. If anyone knows, please feel free to comment. Any suggestions on improving this answer are very welcome, this has been a frustrating dig through documentation, so hopefully this helps point someone else in the right direction.

回答1:

this is how I do it on my end:

    public IQueryable<TEntity> FindNear<TEntity>(string collectionName, Expression<Func<TEntity, object>> field, double longitude, double latitude, double maxDistanceInKm) where TEntity : IEntity
    {
        var collection = database.GetCollection<TEntity>(collectionName);
        var point = GeoJson.Point(GeoJson.Geographic(longitude, latitude));
        var filter = Builders<TEntity>.Filter.Near(field, point, maxDistanceInKm * 1000);
        return collection.Find(filter).ToList().AsQueryable();
    }


回答2:

here's the most convenient way to do geospatial aggregation queries for anybody that's interested:

using MongoDB.Driver;
using MongoDB.Entities;
using System;

namespace StackOverflow
{
    public class Program
    {
        public class Place : Entity
        {
            public string Name { get; set; }
            public DateTime Date { get; set; }
            public Coordinates2D Location { get; set; }
            public double DistanceMeters { get; set; }
        }

        static void Main(string[] args)
        {
            //connect to mongodb
            new DB("test");

            //create a geo2dsphere index
            DB.Index<Place>()
              .Key(x => x.Location, KeyType.Geo2DSphere)
              .Option(x => x.Background = false)
              .Create();

            //create and save a place
            var paris = new Place
            {
                Name = "paris",
                Location = new Coordinates2D(48.8539241, 2.2913515),
                Date = DateTime.UtcNow
            };
            paris.Save();

            var eiffelTower = new Coordinates2D(48.857908, 2.295243);

            //find all places within 1km of eiffel tower.
            var places = DB.GeoNear<Place>(
                              NearCoordinates: eiffelTower,
                              DistanceField: x => x.DistanceMeters,
                              MaxDistance: 1000)
                           .SortByDescending(x=>x.Date)
                           .ToList();
        }
    }
}

it generates the following aggregation pipeline:

{
                "$geoNear": {
                    "near": {
                        "type": "Point",
                        "coordinates": [
                            48.857908,
                            2.295243
                        ]
                    },
                    "distanceField": "DistanceMeters",
                    "spherical": true,
                    "maxDistance": NumberInt("1000")
                }
            },
            {
                "$sort": {
                    "Date": NumberInt("-1")
                }
            }

the above is done using MongoDB.Entities convenience library, of which i'm the author of.



回答3:

This is how I do it. It's very simple and has the bonus of being able to copy and paste your query from the mongo shell into the query variables.

For a simple find, returning 1 document if within 100 metres:

var nearSphereQuery = "{location: { $nearSphere: {$geometry: { type: \"Point\", coordinates: [-1.50,53.50]},  $maxDistance: 100}} }";
var nearSphereBsonDoc = MongoDB.Bson.Serialization.BsonSerializer.Deserialize<BsonDocument>(nearSphereQuery);
var result = points.Find(nearSphereBsonDoc).Limit(1).SingleOrDefault();
if (result != null) {
    // do stuff here
} else {
    // nothing found
}

To get the distance from the search point for the same request:

var geoNearQuery = "{$geoNear: { near: { type: \"Point\", coordinates: [-1.50, 53.50]}, spherical: true, distanceField: \"distance\"} } ,{$match: { distance: { $lt: 100} } }";
var geoNearBsonDoc = MongoDB.Bson.Serialization.BsonSerializer.Deserialize<BsonDocument>(geoNearQuery);
var pipeline = new List<BsonDocument>();
var matchQuery = "{ $match: { distance: { $lt: 10} } }";
var matchBsonDoc = MongoDB.Bson.Serialization.BsonSerializer.Deserialize<BsonDocument>(matchQuery);
var pipeline = new List<BsonDocument>();
pipeline.Add(geoNearBsonDoc);
pipeline.Add(matchBsonDoc);   
pipeline.Add(new BsonDocument { { "$limit", 1 } });
var result = points.Aggregate<BsonDocument>(pipeline).SingleOrDefault();
if (result != null) {
    Double dist = Convert.ToDouble(result["distance"]);
} else {
    // nothing found
}