上一篇文章中，我们学习了Join类型的父子文档，今天继续学习一下嵌套文档，毕竟嵌套文档也是Elasticsearch推荐的，首先我们看下面这条语句

PUT word_document/_doc/1
{"title" : "up","user" : [ {"name" : "honghong","sex" :  "female","numberOfLikes":500},{"name" : "mingming","sex" :  "male","numberOfLikes":50},{"name" : "lanlan","sex" :  "male","numberOfLikes":100}]
}

对于上面这种格式的数据，user就是嵌套对象数组，那么user在Elasticsearch中是怎么存储的呢？如果我们要对嵌套的子对象进行检索，怎么才能检索出我们所需要的数据呢，下面我们就一起来研究下Nested数据类型

环境

• macos 10.14.6
  
• Elasticsearch 8.1
  
• Kibana 8.1
  

Nested

开头我们还是先了解一下，什么是Nested类型，其实就是字面意思，Nested就是嵌套，也就是文章开头user数据类型那种，所以可以看为是一种特殊的Object类型。还是以文章开头的数据为例

PUT word_document/_doc/1
{"title" : "up","user" : [ {"name" : "honghong","sex" :  "female","numberOfLikes":500},{"name" : "mingming","sex" :  "male","numberOfLikes":50},{"name" : "lanlan","sex" :  "male","numberOfLikes":100}]
}

如果我们没有对word_document索引进行显示设置数据类型，在上面这个语句执行之后，Elasticsearch会默认推断数据类型，在Elasticsearch中内容会转换为可能如下的形式，扁平化的处理数据

{  "title":"up",  "user.name":["honghong","mingming","lanlan"],  "user.sex":["male","male","female"],  "user.numberOfLikes":[500,50,100]}

相信大家也看出来了，如果被Elasticsearch转换成上面的这种数据结构之后，我们的搜索结果是会被影响的，假如我们使用如下这个语句进行查询，我们想搜索name是honghong，sex是male，预期结果是没有匹配的文档，但是因为Elasticsearch对上述的结果进行了扁平化的处理，造成了错误的匹配

GET word_document/_search
{"query": {"bool": {"must": [{ "match": { "user.name": "honghong" }},{ "match": { "user.sex":  "male" }}]}}
}

如何避免上述情况的发生呢，那就是使用Elasticsearch提供的Nested数据类型，Nested 数据类型保证了嵌套对象的独立性，也就是让我们可以对嵌套对象的内容进行检索，从而不会发生上述的这种情况

• 首先我们还是以上面文档为例，不过是这次我们优先创建索引，并指定user字段为nested

  PUT word_document
  {"mappings": {"properties": {"title":{"type": "keyword"},"user": {"type": "nested" },"numberOfLikes":{"type": "integer"}}}
  }
  

• 下面加入我们的测试数据，来验证我们的搜索语句

  PUT word_document/_doc/1
  {"title" : "up","user" : [ {"name" : "honghong","sex" :  "female","numberOfLikes":500},{"name" : "mingming","sex" :  "male","numberOfLikes":50},{"name" : "lanlan","sex" :  "male","numberOfLikes":100}]
  }
  PUT word_document/_doc/2
  {"title" : "up","user" : [ {"name" : "honghong","sex" :  "female","numberOfLikes":20},{"name" : "mingming","sex" :  "male","numberOfLikes":30},{"name" : "lanlan","sex" :  "male","numberOfLikes":50}]
  }
  PUT word_document/_doc/3
  {"title" : "up","user" : [ {"name" : "honghong","sex" :  "female","numberOfLikes":50},{"name" : "mingming","sex" :  "male","numberOfLikes":50},{"name" : "lanlan","sex" :  "male","numberOfLikes":50}]
  }
  

• 下面还是刚才那个搜索语句，此时就不会有匹配的文档返回，返回结果为空

  GET word_document/_search
  {"query": {"nested": {"path": "user","query": {"bool": {"must": [{ "match": { "user.name": "honghong" }},{ "match": { "user.sex":  "male" }} ]}}}}
  }
  

• 那么对于嵌套文档我们可以怎么查询呢，那就是指定nested查询类型，使用普通的查询是查询不到的哦，nested查询语句如下所示，此时返回的就是我们

  GET word_document/_search
  {"query": {"nested": {"path": "user","query": {"bool": {"must": [{ "match": { "user.name": "honghong" }},{ "match": { "user.sex":  "female" }} ]}},"inner_hits": { "highlight": {"fields": {"user.name": {}}}}}}
  }
  

• 此外我们还可以根据嵌套对象中的字段进行排序，升序时获取嵌套对象中最小的值最为比较值，降序时获取嵌套对象最大的值作为比较值

  GET word_document/_search
  {"query": {"nested": {"path": "user","query": {"match": {"user.sex": "male"}}}},"sort":[{"user.numberOfLikes": {"order": "asc", "nested": {"path":"user"}}}]
  }
  

  返回如下

  {"took" : 101,"timed_out" : false,"_shards" : {"total" : 1,"successful" : 1,"skipped" : 0,"failed" : 0},"hits" : {"total" : {"value" : 3,"relation" : "eq"},"max_score" : null,"hits" : [{"_index" : "word_document","_id" : "2","_score" : null,"_source" : {"title" : "up","user" : [{"name" : "honghong","sex" : "female","numberOfLikes" : 20},{"name" : "mingming","sex" : "male","numberOfLikes" : 30},{"name" : "lanlan","sex" : "male","numberOfLikes" : 50}]},"sort" : [20]},{"_index" : "word_document","_id" : "1","_score" : null,"_source" : {"title" : "up","user" : [{"name" : "honghong","sex" : "female","numberOfLikes" : 500},{"name" : "mingming","sex" : "male","numberOfLikes" : 50},{"name" : "lanlan","sex" : "male","numberOfLikes" : 100}]},"sort" : [50]},{"_index" : "word_document","_id" : "3","_score" : null,"_source" : {"title" : "up","user" : [{"name" : "honghong","sex" : "female","numberOfLikes" : 50},{"name" : "mingming","sex" : "male","numberOfLikes" : 50},{"name" : "lanlan","sex" : "male","numberOfLikes" : 50}]},"sort" : [50]}]}
  }

• 我们也可以对嵌套对象进行聚合操作，如下我们获取索引中user.name=honghong,user.sex=female的所有文档，聚合统计numberOfLikes的最小值

  GET word_document/_search
  {"query": {"nested": {"path": "user","query": {"bool": {"must": [{"match": {"user.name": "honghong"}},{"match": {"user.sex": "female"}}]}}}},"aggs": {"my_min_value": {"nested": {"path": "user"}, "aggs": {"min_value": {"min": {"field": "user.numberOfLikes"}}}}}
  }
  

• 上面的聚合统计只是对外部的文档过滤，那如果我们有这么一个需求，聚合统计嵌套对象user内容sex=male的最小值，那么我们可以使用如下filter，下面这个语句优先过滤title=up的文档，聚合统计user.sex=male的numberOfLikes最小值

  GET /word_document/_search?size=0
  {"query": {"match": {"title": "up"}},"aggs": {"my_user": {"nested": {"path": "user"},"aggs": {"filter_my_user": {"filter": {"bool": {"filter": [{"match": {"user.sex": "male"}}]}},"aggs": {"min_price": {"min": {"field": "user.numberOfLikes"}}}},"no_filter_my_user":{"min": {"field": "user.numberOfLikes"}}}}}
  }
  

• 最后还有一种就是反向嵌套聚合，通过嵌套对象聚合父文档，返回父文档信息

  首先我们还是先创建一个索引添加几条数据用来测试

  PUT /issues
  {"mappings": {"properties": {"tags": { "type": "keyword" },"comments": {                            "type": "nested","properties": {"username": { "type": "keyword" },"comment": { "type": "text" }}}}}
  }
  PUT /issues/_doc/1
  {"tags":"跳舞","comments":[{"username":"小李","comment":"小李想学跳舞"},{"username":"小红","comment":"小红跳舞很有天赋"}]
  }
  PUT /issues/_doc/2
  {"tags":"唱歌","comments":[{"username":"小李","comment":"小李会唱歌"},{"username":"小李","comment":"小李唱歌有天赋"},{"username":"小红","comment":"小红是歌手"}]
  }
  PUT /issues/_doc/3
  {"tags":"跳舞","comments":[{"username":"小红","comment":"小红会跳舞"},{"username":"小红","comment":"小红是舞神"}]
  }
  PUT /issues/_doc/4
  {"tags":"唱歌","comments":[{"username":"小李","comment":"小李简直就是天生歌手"}]
  }
  PUT /issues/_doc/5
  {"tags":"跳舞","comments":[{"username":"小红","comment":"小红舞姿很美"}]
  }
  

  issues 问题；tags 标签；username 名字；comment 评论；

  下面我们使用反向嵌套聚合父文档，需求如下：

  1、先聚合统计出评论最多的username

  2、在聚合统计username中comment最多的tag

  GET /issues/_search?size=0
  {"query": {"match_all": {}},"aggs": {"comments": {"nested": {"path": "comments"},"aggs": {"top_usernames": {"terms": {"field": "comments.username"},"aggs": {"comment_to_issue": {"reverse_nested": {}, "aggs": {"top_tags_per_comment": {"terms": {"field": "tags"}}}}}}}}}
  }
  

  结果如下，得出结论：小红评论次数最多，评论了5次，小红评论最多的标签是跳舞，有3次

  {"aggregations" : {"comments" : {"doc_count" : 9,"top_usernames" : {"doc_count_error_upper_bound" : 0,"sum_other_doc_count" : 0,"buckets" : [{"key" : "小红","doc_count" : 5,"comment_to_issue" : {"doc_count" : 4,"top_tags_per_comment" : {"doc_count_error_upper_bound" : 0,"sum_other_doc_count" : 0,"buckets" : [{"key" : "跳舞","doc_count" : 3},{"key" : "唱歌","doc_count" : 1}]}}},{"key" : "小李","doc_count" : 4,"comment_to_issue" : {"doc_count" : 3,"top_tags_per_comment" : {"doc_count_error_upper_bound" : 0,"sum_other_doc_count" : 0,"buckets" : [{"key" : "唱歌","doc_count" : 2},{"key" : "跳舞","doc_count" : 1}]}}}]}}}
  }

Nested 支持的参数有哪些

Nested也只是特殊的Object的一种，也是有支持的几种参数

• dynamic: (可选参数) 是否允许在索引 mapping文件未定义字段的情况下对新字段的处理，默认是加入到现有的嵌套对象中( true),还支持 false， strict
  
• properties: (可选参数) 嵌套对象字段内容属性设置
  
• include_in_parent:(可选参数) 默认 false，如果为 true，嵌套对象的字段也会作为普通字段的形式( flat)添加到父文档
  
• include_in_root:(可选参数) 默认 false，如果为 true，嵌套对象的字段也会作为普通字段的形式( flat)添加到根文档
  

Nested 类型的约束

通过前面的学习，我们也知道了nested类型可以作为一个单独的Lucene文档进行所有，当我们有100个嵌套对象的时候我们需要101个文档来存储映射关系，一个用于父文档，一个用于嵌套文档，所以这一部分的开销，ELasticsearch来通过一下设置进行了约束

• index.mapping.nested_fields.limit

  一个索引中，嵌套类型字段(nested)最多存在多个限制，默认50个，如我们上面的例子中，也就是只占用了一个

• index.mapping.nested_objects.limit

  一个索引中，单个嵌套类型字段包含的嵌套JSON对象的最大数量，默认10000

总结

通过上面的学习实践，我们可以知道Nested嵌套类型是Elasticsearch推荐的相对于Join类型，并且Nested可以实现查询，聚合，排序等，基本满足了工作的需要。好了，到这就结束吧，有什么需要深入了解的，留言哦，也可以去官网查看，毕竟官网还是一手资料，博主的也只能算是入门启蒙笔记，实践起来吧，加油！

Join 字段的详解可以参考博主的这一篇文章哦