用于地理搜索的 pymongo.cursor 输出缓慢

Question

我对一个充满多边形的集合执行地理搜索。查询本身非常快，但需要 3 多分钟才能得到结果。我需要一个解决方案来加快多边形名称的输出。

我尝试只包含名称，排除几何形状，根本没有指定任何内容，但效果相同。

输出：

0:00:00.000061
Result of Query
<class 'pymongo.cursor.Cursor'>
{'_id': ObjectId('5e70ab3b160bdb5ebde5da7f'), 'name': 'Hessen'}
{'_id': ObjectId('5e70ab65160bdb5ebde5e0d4'), 'name': 'Hessen/Rheingau'}
{'_id': ObjectId('5e70ab7a160bdb5ebde5e38e'), 'name': 'Hessen'}
{'_id': ObjectId('5e70b707faa4a158f866f6fb'), 'name': 'Frankfurt am Main'}
0:03:11.020437

from pymongo import MongoClient
import json
from datetime import datetime

client = MongoClient()
client = MongoClient('localhost', 27017)
db = client['GEO']
geo_collection = db['polygons']

starttime = datetime.now()
search_results = geo_collection.find({"geometry":
                 {"$geoIntersects":
                     {"$geometry":{ "type" : "Point",
                          "coordinates" :  [8.636274, 50.145938]}
                      }
                  }
             },{"geometry":0})

endtime = datetime.now()
print(endtime-starttime)

print("Result of Query")
print(type(search_results))
for result in search_results:
    print(result)

endtime = datetime.now()
print(endtime-starttime)

> db.polygons.explain().find({"geometry":{"$geoIntersects":{"$geometry":{ "type" : "Point","coordinates":[8.636274, 50.145938]}}}},{"name":1})
{
    "queryPlanner" : {
        "plannerVersion" : 1,
        "namespace" : "GEO.polygons",
        "indexFilterSet" : false,
        "parsedQuery" : {
            "geometry" : {
                "$geoIntersects" : {
                    "$geometry" : {
                        "type" : "Point",
                        "coordinates" : [
                            8.636274,
                            50.145938
                        ]
                    }
                }
            }
        },
        "queryHash" : "D3B4DDC2",
        "planCacheKey" : "45C3B08C",
        "winningPlan" : {
            "stage" : "PROJECTION_SIMPLE",
            "transformBy" : {
                "name" : 1
            },
            "inputStage" : {
                "stage" : "COLLSCAN",
                "filter" : {
                    "geometry" : {
                        "$geoIntersects" : {
                            "$geometry" : {
                                "type" : "Point",
                                "coordinates" : [
                                    8.636274,
                                    50.145938
                                ]
                            }
                        }
                    }
                },
                "direction" : "forward"
            }
        },
        "rejectedPlans" : [ ]
    },
    "serverInfo" : {
        "host" : "DTO-MB-007.local",
        "port" : 27017,
        "version" : "4.2.3",
        "gitVersion" : "6874650b362138df74be53d366bbefc321ea32d4"
    },
    "ok" : 1
}

Answer 1

“2d”索引支持传统的长/纬度坐标对，而不是像 Polygon 这样的 GeoJSON 对象。

2dsphere index 确实支持这些。 geometry 字段似乎是格式正确的 GeoJSON 对象，因此您应该能够创建如下索引：

db.polygons.createIndex({"geometry":"2dsphere"})

该索引应支持所注明的查询。