根据索引文档的地理坐标来进行搜索,Elasticsearch 也能够处理这样的搜索。——空间搜索
一、为空间搜索准备映射
PUT my_space_test
{
"mappings": {
"poi": {
"properties": {
"name": {
"type": "string"
},
"locationpoint": {
"type":"geo_point" //任意的地理坐标
},
"locationshape": {
"type": "geo_shape" //任意的地理形状
}
}
}
}
}
二、批量添加数据
POST my_space_test/poi/_bulk
{"index":{"_id":1}}
{"name":"New York","locationpoint":"40.664167, -73.938611","locationshape":{"type":"polygon","coordinates":[[[4.8833,52.38617],[4.87463,52.37254],[4.87875,52.36369],[4.8833,52.38617]]]}}
{"index":{"_id":2}}
{"name":"London","locationpoint":[-0.1275,51.5072222],"locationshape":{"type":"polygon","coordinates":[[[0,0],[4.87463,52.37254],[4.87875,52.36369],[0,0]]]}}
{"index":{"_id":3}}
{"name":"Moscow","locationpoint":{"lat":55.75,"lon":37.616667},"locationshape":{"type":"polygon","coordinates":[[[22,22],[4.87463,52.37254],[4.87875,52.36369],[22,22]]]}}
{"index":{"_id":4}}
{"name":"Sydney","locationpoint":"-33.859972, 151.211111","locationshape":{"type":"polygon","coordinates":[[[4.8833,52.38617],[4.87463,52.37254],[4.87875,52.36369],[4.8833,52.38617]]]}}
{"index":{"_id":5}}
{"name":"Sydney","locationpoint":"eycs0p8ukc7v","locationshape":{"type":"polygon","coordinates":[[[4.8833,52.38617],[4.87463,52.37254],[4.87875,52.36369],[4.8833,52.38617]]]}}
仔细观看locationpoint字段可以看到坐标可以使用多种形式来赋值,可以使用字符串、数组(只能包含两个数值)、一个对象、地理散列值等来提供经纬度。具体每种方式可能略有不同,具体使用再查相关资料。
再来看一下,locationshape,其形式就更加多样了,可以是一个点 ,即为一组数值对 [ 经度,维度 ] ,也可以是一个框 [ [左,上], [右,下] ],还可以是多边形,但是必须保证第一个坐标和最后一个坐标是相同的,从而保证是一个闭合的图形。[ [ [1,1],[2,2], [3,4] ,[1,1] ] ] ,可以发现多边形的定义中其可以是多个多边形,是一个可扩展的数组。
三、查询方式
3.1 基于距离排序
GET my_space_test/poi/_search
{
"query": {
"match_all": {}
},
"sort": [
{
"_geo_distance": {
"locationpoint": {
"lat": 48.8567,
"lon": 2.3508
},
"unit": "km",
"order": "asc"
}
}
]
}
GET my_space_test/poi/_search
{
"query": {
"match_all": {}
},
"sort": [
{
"_geo_distance":<{
"locationpoint": [ //或者是:"locationpoint":" 48.8567,2.3508"
2.3508,
48.8567
],
"unit": "km",
"order": "asc"
}
}
]
}
以上查询的结果是一样的(注意数组和字符串坐标的位置顺序是不同的)通过距离坐标点 [ 2.3508,48.8567 ]的大小来对查询文档进行排序。这在实际搜索中非常有用,可以返回临近的一些坐标点。
3.2 边界框过滤(获得包含在指定区域内文档)
"query": {
"filtered": {
"filter": {
"bool": {
"should": [
{
"geo_bounding_box": {
"locationpoint": {
"top_left": "52.4796,-1.903",
"bottom_right": "48.8567,2.3508"
}
}
},
{
"geo_distance":{
"distance": 500,
"distance_unit": "km",
"locationpoint": "48.8567,2.3508"
}
}]}}}}
示意图 1
返回的文档就像是包含在矩形和圆形中的蓝色点,红色点用来确定边框,红色线段确定距离范围。在图形之外的点就被过滤掉了。以上都是针对类型为geo_point
下面我们来看一下geo_shape类型是如何使用的?
"query": {
"filtered": {
"filter": {
"bool": {
"should": [
{
"geo_shape": {
"locationshape": {
"indexed_shape": { //使用已经索引的形状
"index": "my_space_test",
"type": "poi",
"id": "4",
"path": "locationshape"
},
"relation": "within"
}
}
},
{
"geo_shape": {
"locationshape": { //自定义形状——圆
"shape": {
"type": "circle",
"radius": "1km",
"coordinates": [
-45,
45
]
},
"relation": "within"
}
}
},{
"geo_shape": {
"locationshape": {
"shape": {
"type": "envelope", //自定义的形状包络线,即:box(矩形)
"coordinates": [
[
-45,
45
],
[
45,
-45
]
]
},
"relation": "within"
}
}
},{
"geo_shape": {
"locationshape": { //自定义的多边形,一定要注意,多边形的定义是包含在一个数组中的,是一个可扩展的数组
"shape": {
"type": "polygon",
"coordinates": [[[1,1],[2,3],[4,2],[1,1]]]
},
"relation": "within"
}
}
}
]
}
}
}
}
首先来说,过滤查询的字段locationshap 中包含多种形状类型,有点、包络线、多边形、甚至说多个多边形
以上的查询是看那些形状位于所查询的形状之内。
我们再来个示意图吧,这样好理解一些。
示意图2
比如说,我们可以定义一个中国的多边形,然后查找那些城市是位于中国的,这些城市也可以是多边形,当然也可以用一个点来定义,通过这样的过滤方式都能够准确的找到。
怎么样?ES是不是很炫?革命尚未成功,同志仍需努力!坚持你才能看到最美的风景,即便一路上会有荆棘。接下来看看,空间搜索对应的高亮和聚合。待续……