1.背景介绍
我们在使用Elasticsearch做搜索引擎的时候有可能会遇到跨domain查询的场景,比如做一个学生课程管理系统,搜一个学生的名字,像知道该学生的选课情况。
当然解决问题的方法有很多,我可以搜学生,然后去db查找学生关联的选课,就可以查到所有的课程,有时候数据量不是很大,并且我的索引只有一个课程维度的时候,就需要使用嵌套类型来解决这类问题。本文使用es和kibina来操作实例,因为基于中文的实例,还使用到了ik分词器,具体可以参考:
2.对象类型
Elasticsearch支持对象类型的存储,我们可以把一个对象数组存到某个document的字段内,比如一个课程作为一个document,那么这个课程可以建立一个students字段,存储该课程下的学生object数组。
在Elasticsearch中,新建一个如下的class_test索引,其中student作为一个object数组类型。
- PUT /class_test
- {
- "mappings":{
- "class_test": {
- "properties": {
- "id": {
- "type": "keyword"
- },
- "name": {
- "analyzer": "ik_max_word",
- "type": "text"
- },
- "type":{
- "type":"keyword"
- },
- "student":{
- "properties": {
- "name":{
- "analyzer": "ik_max_word",
- "type": "text"
- },
- "id":{
- "type":"keyword"
- }
- }
- }
- }
- }
- },
- "settings":{
- "index": {
- "refresh_interval": "1s",
- "number_of_shards": 5,
- "max_result_window": "10000000",
- "mapper": {
- "dynamic": "false"
- },
- "number_of_replicas": 0
- }
- }
- }复制代码
往class_test放入一下数据,现在索引里面一共有两条数据
- {
- "took" : 1,
- "timed_out" : false,
- "_shards" : {
- "total" : 5,
- "successful" : 5,
- "skipped" : 0,
- "failed" : 0
- },
- "hits" : {
- "total" : 2,
- "max_score" : 1.0,
- "hits" : [
- {
- "_index" : "class_test",
- "_type" : "class_test",
- "_id" : "ijfJ5GoBJeNZPNCWykLR",
- "_score" : 1.0,
- "_source" : {
- "id" : "1",
- "name" : "数学课",
- "student" : [
- {
- "id" : "1",
- "name" : "张三"
- },
- {
- "id" : "2",
- "name" : "李四"
- }
- ]
- }
- },
- {
- "_index" : "class_test",
- "_type" : "class_test",
- "_id" : "Q9NxGGsBa-TqHCWqAaM4",
- "_score" : 1.0,
- "_source" : {
- "id" : "2",
- "name" : "语文",
- "student" : [
- {
- "id" : "3",
- "name" : "杰克"
- },
- {
- "id" : "4",
- "name" : "玛丽"
- }
- ]
- }
- }
- ]
- }
- }复制代码
接下来,我们可以使用查询语句对索引进行查询。当我们查询id为1的学生参见的课程的时候,可以查到数学课。
- GET /class_test/class_test/_search
- {
- "query": {
- "bool": {
- "must": [
- {
- "match": {
- "student.id": "1"
- }
- }
- ]
- }
- }
- }复制代码
- {
- "took" : 0,
- "timed_out" : false,
- "_shards" : {
- "total" : 5,
- "successful" : 5,
- "skipped" : 0,
- "failed" : 0
- },
- "hits" : {
- "total" : 1,
- "max_score" : 0.2876821,
- "hits" : [
- {
- "_index" : "class_test",
- "_type" : "class_test",
- "_id" : "ijfJ5GoBJeNZPNCWykLR",
- "_score" : 0.2876821,
- "_source" : {
- "id" : "1",
- "name" : "数学课",
- "student" : [
- {
- "id" : "1",
- "name" : "张三"
- },
- {
- "id" : "2",
- "name" : "李四"
- }
- ]
- }
- }
- ]
- }
- }
- 复制代码
当我们查名字叫张三的学生参加的课程的时候,也能查到数学课。
- GET /class_test/class_test/_search
- {
- "query": {
- "bool": {
- "must": [
- {
- "match": {
- "student.name": "张三"
- }
- }
- ]
- }
- }
- }复制代码
- {
- "took" : 4,
- "timed_out" : false,
- "_shards" : {
- "total" : 5,
- "successful" : 5,
- "skipped" : 0,
- "failed" : 0
- },
- "hits" : {
- "total" : 1,
- "max_score" : 0.5753642,
- "hits" : [
- {
- "_index" : "class_test",
- "_type" : "class_test",
- "_id" : "ijfJ5GoBJeNZPNCWykLR",
- "_score" : 0.5753642,
- "_source" : {
- "id" : "1",
- "name" : "数学课",
- "student" : [
- {
- "id" : "1",
- "name" : "张三"
- },
- {
- "id" : "2",
- "name" : "李四"
- }
- ]
- }
- }
- ]
- }
- }复制代码
但是当我们查询id为1并且名字叫李四的学生参加的课程时
- GET /class_test/class_test/_search
- {
- "query": {
- "bool": {
- "must": [
- {
- "match": {
- "student.name": "李四"
- }
- },
- {
- "match": {
- "student.id": "1"
- }
- }
- ]
- }
- }
- }复制代码
- {
- "took" : 6,
- "timed_out" : false,
- "_shards" : {
- "total" : 5,
- "successful" : 5,
- "skipped" : 0,
- "failed" : 0
- },
- "hits" : {
- "total" : 1,
- "max_score" : 0.8630463,
- "hits" : [
- {
- "_index" : "class_test",
- "_type" : "class_test",
- "_id" : "ijfJ5GoBJeNZPNCWykLR",
- "_score" : 0.8630463,
- "_source" : {
- "id" : "1",
- "name" : "数学课",
- "student" : [
- {
- "id" : "1",
- "name" : "张三"
- },
- {
- "id" : "2",
- "name" : "李四"
- }
- ]
- }
- }
- ]
- }
- }
- 复制代码
我们发现,出来的结果也是数学课,这就有点奇怪,因为并没有一个id为1并且名字是李四的学生,那就不应该有这么课。这是怎么回事?原来在es内部,object数组类型会被打平,简单来说我们输入的数组,实际存储的类型是:
- "student.id":[1,2],
- "student.name":[张三,李四]复制代码
所以倒排索引的建立,也是按照这种打平的逻辑。这个时候我们可以借助Elasticsearch内的嵌套类型来解决问题。
3.Nested类型
和2中类似的,我们需要建一个测试索引,名字为class,不同的是student有了type字段,为 "type":"nested"。
- PUT /class
- {
- "mappings":{
- "class": {
- "properties": {
- "id": {
- "type": "keyword"
- },
- "name": {
- "analyzer": "ik_max_word",
- "type": "text"
- },
- "type":{
- "type":"keyword"
- },
- "student":{
- "type":"nested",
- "properties": {
- "name":{
- "analyzer": "ik_max_word",
- "type": "text"
- },
- "id":{
- "type":"keyword"
- }
- }
-
- }
- }
- }
- },
- "settings":{
- "index": {
- "refresh_interval": "1s",
- "number_of_shards": 5,
- "max_result_window": "10000000",
- "mapper": {
- "dynamic": "false"
- },
- "number_of_replicas": 0
- }
- }
- }复制代码
我们导入相同的数据,然后用搜索id为1并且名字为李四的学生的课程,这个时候我们看到搜索结果为空:
- GET /class/class/_search
- {
- "query": {
- "bool": {
- "must": [
- {
- "nested": {
- "path": "student",
- "query": {
- "bool": {"must": [
- {
- "match": {
- "student.name": "李四"
- }
- },
- {
- "match": {
- "student.id": "1"
- }
- }
- ]}
- }
- }
- }
- ]
- }
- }
- }复制代码
- {
- "took" : 3,
- "timed_out" : false,
- "_shards" : {
- "total" : 5,
- "successful" : 5,
- "skipped" : 0,
- "failed" : 0
- },
- "hits" : {
- "total" : 0,
- "max_score" : null,
- "hits" : [ ]
- }
- }
- 复制代码
4.其他方式
其实解决这种跨domain的搜索还有一些其他方式,对于嵌套类型,其实是非常消耗Elasticsearch的性能的,我们可以选择将需要搜索字段的值打平存一个字段,或者对学生单独建立一个索引,然后去学生-班级映射关系表查询班级。这一块后面有机会再做介绍。