第1位高富帅,来自南京西路东路地铁站
POST my_index/my_type/2 { "group":"白富美", "id": "2234", "sex": 2, "attribute": "eat,dog,boys", "birthday": "1989-12-10", "lat_lng": "31.2433470000,121.5087220000" }第2位白富美来自陆家嘴
PUT my_index/my_type/3 { "group":"小妹妹", "id": "3234", "sex": 2, "attribute": "eat dog boy flowers", "birthday": "2010-12-10", "lat_lng": "31.2257000000,121.5508340000" }第3位小妹妹来自世纪大道
PUT my_index/my_type/4 { "group":"空姐", "id": "4234", "sex": 2, "attribute": "eat,dog,girl", "birthday": "1995-12-10", "lat_lng": "31.1573860000,121.8150200000" }第4位空姐,来自浦东机场
看下刚刚我们建的索引的mapping长啥样
GET my_index/_mapping/my_typeresponse:
{ "my_index": { "mappings": { "my_type": { "properties": { "attribute": { "type": "string" }, "birthday": { "type": "date", "format": "strict_date_optional_time||epoch_millis" }, "group": { "type": "string" }, "id": { "type": "string" }, "lat_lng": { "type": "string" }, "sex": { "type": "long" } } } } } }很明显,mapping 定义了每个field的数据类型(用途之一),es很聪明,能自动确定类型:
“1900-12-10” -> date 1 -> long
然而,有些field要让它完全猜对我们的心思还是有些强人所难,比如:
“1234” -> string “31.2427760000,121.4903420000” -> striing
我其实希望
“1234” - > int “31.2427760000,121.4903420000” -> (维度,经度)
后面会提到怎样修改type,不过在此之前,先了解下es有哪些type
• String:string • Whole number: byte, short, integer, long(默认) • Floating-point:float,double • Boolean:boolean • Date:date • lat/lon points:geo_point
所有的type见:Field datatypes
要改变field的类型,必须先删掉之前的索引!
DELETE /my_index再重新建
PUT /my_index { "mappings": { "my_type": { "properties": { "id": { "type": "string" }, "birthday": { "type": "date" }, "sex": { "type": "short" }, "attribute": { "type": "string" }, "lat_lng": { "type": "geo_point" } } } } }检查下
GET my_index/_mapping/my_type { "my_index": { "mappings": { "my_type": { "properties": { "attribute": { "type": "string" }, "birthday": { "type": "date", "format": "strict_date_optional_time||epoch_millis" }, "id": { "type": "string" }, "lat_lng": { "type": "geo_point" }, "sex": { "type": "short" } } } } } }再将之前的4个doc全塞进去! 很好,type就是我们想要的啦,有啥用,目前的app很多有附近搜索,我们也来小试牛刀下
lat_lng:搜索的中心,我这里用的是人民广场 distance:搜索半径,我这里设为4km,单位可以是m,有兴趣的可以了解下geohash,就大概知道为啥能这么快实现啦
当然,还可以指定区域搜索,更多精彩内容见:Geo Distance Query,Geo Location and Search
ref:Analysis and Analyzers 上一节已经稍稍讲过analyzer,总之,建索引的时候,es会按每个field配置的analyzer分析field值,用来建倒排索引(Inverted Index),搜索的时候,也会按搜索字段的analyzer分析查询值。 和type类似的思路,先了解下es有哪些analyzer,接着指定analyzer。
介绍2个简单的,自己运行理解下吧 - whitespace(空格)
GET /_analyze?analyzer=whitespace { "text":"full-text books, tired sleeping" } english GET /_analyze?analyzer=english { "text":"full-text books, tired sleeping" }分词,词干化后剩下:full,text,book,tire,sleep 其它Built-in Analyzers analyzer,自定义analyzer见:Analyzers
同样要先删掉之前的索引
DELETE /my_index PUT /my_index { "mappings": { "my_type": { "properties": { "id": { "index": "no", "type": "string" }, "birthday": { "index": "not_analyzed", "type": "date" }, "sex": { "index": "not_analyzed", "type": "short" }, "attribute": { "index": "analyzed", "analyzer": "whitespace", "type": "string" }, "lat_lng": { "type": "geo_point" } } } } }index:控制fiel是否被索引,是否要分析
no:指定field不参与建索引,当然也无法搜索该字段 analyzed:分析字段(缺省时,默认analyzed) not_analyzed:不分析字段
analyzer:控制怎样被索引(缺省时,默认 standard )
补充一个可以搜索所有字段的内容,_all field,要去找房子啦o(╯□╰)o,自己看看吧
"_all": { "enabled": false }