elastic search使用案例

本文只是写一些常用es命令，这里不用任何客户端，只用 9200/_plugin/head/ 那个端口网页就可以，本文以sql的角度描述这些命令，因为程序员对于sql更加了解。

注意es其实一个idnex只能有一个type，如果一个index做了多个type，坑。。。太多了，首先 where不能用，然后 聚合也不能用。至少在es2.*是这样的，这个坑你可以搜索下“_type Field Not Available for Aggregation #5634”

1. 查询所有
关键字：match_all
等同SQL

select * from item.ic_spu;
{
"query": {
"match_all": {}
}
}
2. 根据某个字段查询 – 默认匹配
关键字：match

等同SQL

select * from item.ic_spu where id = 20001137 ;
{
"query": {
"match": {
"id": "20001137"
}
}
}
{
"query": {
"match": {
"tags.tagId": "30000"
}
}
}
 

{
"query": {
"match": {
"title": "黄连丸"
}
}
}

这三种查询都会在item_repository/item库中匹配出的结果示例

{
"_index": "item_repository",
"_type": "item",
"_id": "20001137",
"_version": 101,
"found": true,
"_source": {
"seller": {
"name": "好药师官方旗舰店",
"alias": "好药师大药房",
"id": 989810607
},
"modifiedTime": "2015-12-31 11:31:04.000",
"img": "T1E3YlByJT1RCvBVdK|T14ddTBsZT1RCvBVdK|T1jdCTBjxT1RCvBVdK|T1fJETBjWT1RCvBVdK|T1IJdTBjAT1RCvBVdK",
"skus": [{
"hcodePrice": 0,
"origPrice": 1500,
"outerId": "2115",
"id": 20001513,
"price": 1528,
"status": 2
}],
"title": "清胃黄连丸",
"type": 20,
"tags": [{
"code": "webFronts2",
"tagName": "五官科疾病",
"tagId": 30017,
"domain": "mall_sales"
},
{
"code": "o2oFronts",
"tagName": "感冒发烧",
"tagId": 40002,
"domain": "mall_sales"
},
{
"code": "webFronts1",
"tagName": "中西药品",
"tagId": 30000,
"domain": "mall_sales"
},
{
"code": "webFronts3",
"tagName": "口腔/咽喉",
"tagId": 30132,
"domain": "mall_sales"
},
{
"code": "webFronts",
"tagName": "前台根节点",
"tagId": 2,
"domain": "mall_sales"
},
{
"code": "normal",
"tagName": "normal",
"tagId": 20839,
"domain": "mall_sales"
}],
"timeStamp": "2016-12-27 16:00:30.758",
"subTitle": "",
"createTime": "2015-12-31 11:31:04.000",
"indexId": "20001137",
"subType": 15,
"id": 20001137,
"status": 2
}
}

3.根据某个字段查询 – 绝对匹配
关键字：match + phrase
JQuery示例
根据id查询

{
"query": {
"match": {
"id": {
"query": "20001137",
"type": "phrase"
}
}
}
}

注意：类似 id 等数值类型的属性，默认匹配和绝对匹配结果是相同的
对于文本类的属性，没有做分词（not_analyzed）的属性，绝对匹配时用属性完整的值可以匹配出预期结果，例如item库中的tags.domain

如果设置了analyzer，用属性完整的值去做绝对匹配可能会无结果，如item库中的title属性

原因：这种情况下的匹配，是用属性完整的值去匹配索引中被分词后的属性值，自然就可能匹配不上。
例如某spu名称为“感冒颗粒”，那么索引中实际上存储的是分词后的“感冒”和“颗粒”，此时用“感冒颗粒”去匹配，当然无法绝对匹配到“感冒”或者“颗粒”
根据文本查询（针对无analyzer的属性）
等同SQL

select * from item.ic_tag where domain = "mall_sales" ;
{
"query": {
"match": {
"tags.domain": {
"query": "mall_sales",
"type": "phrase"
}
}
}
}

这些查询都会在item_repository/item库中匹配出的结果示例

{
"_index": "item_repository",
"_type": "item",
"_id": "20001137",
"_version": 101,
"found": true,
"_source": {
"seller": {
"name": "好药师官方旗舰店",
"alias": "好药师大药房",
"id": 989810607
},
"modifiedTime": "2015-12-31 11:31:04.000",
"img": "T1E3YlByJT1RCvBVdK|T14ddTBsZT1RCvBVdK|T1jdCTBjxT1RCvBVdK|T1fJETBjWT1RCvBVdK|T1IJdTBjAT1RCvBVdK",
"skus": [{
"hcodePrice": 0,
"origPrice": 1500,
"outerId": "2115",
"id": 20001513,
"price": 1528,
"status": 2
}],
"title": "清胃黄连丸",
"type": 20,
"tags": [{
"code": "webFronts2",
"tagName": "五官科疾病",
"tagId": 30017,
"domain": "mall_sales"
},
{
"code": "o2oFronts",
"tagName": "感冒发烧",
"tagId": 40002,
"domain": "mall_sales"
},
{
"code": "webFronts1",
"tagName": "中西药品",
"tagId": 30000,
"domain": "mall_sales"
},
{
"code": "webFronts3",
"tagName": "口腔/咽喉",
"tagId": 30132,
"domain": "mall_sales"
},
{
"code": "webFronts",
"tagName": "前台根节点",
"tagId": 2,
"domain": "mall_sales"
},
{
"code": "normal",
"tagName": "normal",
"tagId": 20839,
"domain": "mall_sales"
}],
"timeStamp": "2016-12-27 16:00:30.758",
"subTitle": "",
"createTime": "2015-12-31 11:31:04.000",
"indexId": "20001137",
"subType": 15,
"id": 20001137,
"status": 2
}
}

4. 根据某个字段查询 – 通配符匹配
关键字：wildcard
？代表任意1个字符，*代表任意多个字符
JQuery示例
注意：类似 id 等数值类型的属性，也可使用通配符匹配，如同sql中的like查询
对于文本类的属性，纯粹的通配符匹配（即：预期返回以查询keyword做like查询的结果）只适用于没有做分词（not_analyzed）的属性，例如item库中的tags.domain

如果设置了analyzer，通配符匹配也要结合分词去考虑，实际上只能用分词加通配符匹配出预期结果，例如“ *白药* ”可以匹配出“云南白药膏”，但是“ *白药牙* ”就匹配不出。

查看索引分词结果可用 GET /item_repository/_analyze?analyzer=shennong-max&pretty=true&text=(云南白药膏)
根据文本查询（针对无analyzer的属性）
等同SQL

select * from item.ic_tag where domain like "?al%" ;
{
"query": {
"wildcard": {
"tags.domain": "?al*"
}
}
}

这些查询都会在item_repository/item库中匹配出的结果示例
{
"_index": "item_repository",
"_type": "item",
"_id": "20001137",
"_version": 101,
"found": true,
"_source": {
"seller": {
"name": "好药师官方旗舰店",
"alias": "好药师大药房",
"id": 989810607
},
"modifiedTime": "2015-12-31 11:31:04.000",
"img": "T1E3YlByJT1RCvBVdK|T14ddTBsZT1RCvBVdK|T1jdCTBjxT1RCvBVdK|T1fJETBjWT1RCvBVdK|T1IJdTBjAT1RCvBVdK",
"skus": [{
"hcodePrice": 0,
"origPrice": 1500,
"outerId": "2115",
"id": 20001513,
"price": 1528,
"status": 2
}],
"title": "清胃黄连丸",
"type": 20,
"tags": [{
"code": "webFronts2",
"tagName": "五官科疾病",
"tagId": 30017,
"domain": "mall_sales"
},
{
"code": "o2oFronts",
"tagName": "感冒发烧",
"tagId": 40002,
"domain": "mall_sales"
},
{
"code": "webFronts1",
"tagName": "中西药品",
"tagId": 30000,
"domain": "mall_sales"
},
{
"code": "webFronts3",
"tagName": "口腔/咽喉",
"tagId": 30132,
"domain": "mall_sales"
},
{
"code": "webFronts",
"tagName": "前台根节点",
"tagId": 2,
"domain": "mall_sales"
},
{
"code": "normal",
"tagName": "normal",
"tagId": 20839,
"domain": "mall_sales"
}],
"timeStamp": "2016-12-27 16:00:30.758",
"subTitle": "",
"createTime": "2015-12-31 11:31:04.000",
"indexId": "20001137",
"subType": 15,
"id": 20001137,
"status": 2
}
}
5.根据某个字段查询 – 前缀匹配
关键字：prefix
JQuery示例
注意：可以把前缀匹配理解为一种特殊的通配符匹配
对于文本类的属性，前缀匹配和通配符匹配一样需要考虑analyzed的影响
根据文本查询（针对无analyzer的属性）
等同SQLselect * from item.ic_tag where domain like "mal%" ;
{
"query": {
"prefix": {
"tags.domain": "mal"
}
}
}

6. 根据某个字段查询 – 范围匹配
关键字：range
JQuery示例
等同SQL
select * from item.ic_spu where id >= 20001137 and id <= 20001140;
注意：范围匹配对数值类型和时间类型的属性都适用

{
"query": {
"range": {
"id": {
"include_lower": true,
"from": "20001137",
"include_upper": true,
"to": "20001140"
}
}
}
}
{
"query": {
"range": {
"modifiedTime": {
"include_lower": false,
"from": "2015-12-31 11:31:04.000",
"include_upper": true,
"to": "2015-12-31 11:31:15.000"
}
}
}
}
7. AND 查询
关键字：bool + must
JQuery示例

{
"query": {
"bool": {
"must": [
{
"match": {
"type": 20
}
},
{
"match": {
"subType": 15
}
},
{
"query": {
"wildcard": {
"title": "*感冒*"
}
}
}
]
}
}
}

8. OR 查询
关键字：bool + should + minimum_should_match
JQuery示例
注意："minimum_should_match": 1 可以不写，表示should结构体中至少得满足一个条件，即通常意义上的or，这也是默认值。
当然，对于should结构体中包括>2个条件的情况，可以设m来构造必须满足其中m个条件的广义OR查询
例如，下面的JQuery，如果设成2 其实也等同于AND查询

{
"query": {
"bool": {
"should": [
{
"match": {
"type": 20
}
},
{
"match": {
"tags.domain": {
"query": "mall_admin",
"type": "phrase"
}
}
}
],
"minimum_should_match": 1
}
}

9. NOT 查询
关键字：bool + must_not
JQuery示例
根据id查询
等同SQL

select * from item.ic_spu where type != 20 ;
{
"query": {
"bool": {
"must_not": [
{
"match": {
"type": 20
}
}
]
}
}
}

10. 分页与排序
关键字：from + size + sort
JQuery示例
根据id查询
等同SQL 

select * from item.ic_spu order by gmt_modified limit 1,5;

{
"size": "5",
"from": "1",
"sort": [
{
"modifiedTime": "asc"
}
]
}