Elasticsearch 搜索基本操作

2021年初报名了阿里云举办的elasticsearch 百人大作战，共同完成《ELK操作手册》的编写，有幸参与到了基础能力编写的部分-search基本操作，现整理部分内容展现给大家参考学习。

业务背景

在2B行业，对商品的搜索展示是有一定业务要求的，例如：存在合作关系的买家和供应商才能看到供应商店铺的商品，不存在合作关系的买家则不展示商品，另外，有些商品对客户甲展示一种价格，对另外一些客户则展示另外一种价格，从而区分不同的会员、分组对商品价格的区别。一句话总结：2B行业的商品销售具有一定封闭性、特殊性。后续例子均在此背景下展开描述，以方便大家更加贴近业务场景来熟悉elastic search 对文档、索引、查询的一系列操作。

定义mapping

商品字段描述如下：

goodsName: 商品名称
skuCode：商品sku编码
brandName：商品品牌名称
channelType：渠道类型
shopCode：店铺编码
publicPrice：售卖价格（基础价，对所有人开放价格）
closeUserCode：封闭会员编码
groupPrice：分组价格，其中使用嵌套类型存储
boxLevelPrice：分组价格
level：分组级别

定义商品mapping

PUT my_goods_20210423
{
  "settings": {
    "index": {
      "number_of_shards": 1,
      "number_of_replicas": 1
    }
  },
  "mappings": {
    "properties": {
      "goodsName": {
        "type": "text",
        "analyzer": "ik_smart"
      },
      "skuCode": {
        "type": "keyword"
      },
      "brandName": {
        "type": "keyword"
      },
      "channelType": {
        "type": "keyword"
      },
      "shopCode": {
        "type": "keyword"
      },
      "publicPrice": {
        "type": "float"
      },
      "closeUserCode": {
        "type": "text",
        "analyzer": "standard"
      },
      "boostValue": {
        "type": "keyword"
      },
      "groupPrice": {
        "type": "nested",
        "properties": {
          "boxLevelPrice": {
            "type": "float"
          },
          "level": {
            "type": "text"
          }
        }
      }
    }
  }
}

操作文档

主要涉及以下几个核心功能

document.png

1、新增

对文档的新增操作支持以下类型

PUT /<target>/_doc/<_id>

POST /<target>/_doc/

PUT /<target>/_create/<_id>

POST /<target>/_create/<_id>

以 POST /<target>/_create/<_id>为例，以下将创建文档ID为1的商品信息：

POST /my_goods_20210423/_create/1
{
    "goodsName":"苹果 51英寸 4K超高清",
    "skuCode":"skuCode1",
    "brandName":"苹果",
    "closeUserCode":[
        "0"
    ],
    "channelType":"cloudPlatform",
    "shopCode":"sc00001",
    "publicPrice":"8188.88",
    "groupPrice":null,
    "boxPrice":null,
    "boostValue":1.8
}

ES支持批量插入，_bulk桶插入

POST my_goods_20210423/_bulk
{"index":{"_id":3}}
{"goodsName":"苹果UA55RU7520JXXZ 53英寸 4K高清","skuCode":"skuCode3","brandName":"美国苹果","closeUserCode":["0"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8388.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["htd003","uc004"],"boxPriceDetail":4388.88},{"boxType":"box2","boxUserCode":["uc005","uc0010"],"boxPriceDetail":5388.88}],"boostValue":1.2}
{"index":{"_id":4}}
{"goodsName":"山东苹果UA55RU7520JXXZ 苹果54英寸 5K超高清","skuCode":"skuCode4","brandName":"山东苹果","closeUserCode":["uc001","uc002","uc003"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8488.88","groupPrice":[{"level":"level1","boxLevelPrice":"2488.88"},{"level":"level2","boxLevelPrice":"3488.88"}],"boxPrice":[{"boxType":"box1","boxUserCode":["uc004","uc005","uc006","uc001"],"boxPriceDetail":4488.88},{"boxType":"box2","boxUserCode":["htd007","htd008","htd009","uc0010"],"boxPriceDetail":5488.88}],"boostValue":1.2}
{"index":{"_id":5}}
{"goodsName":"苹果UA55R苹果U7苹果520JXXZ 55英寸 5K超高清","skuCode":"skuCode5","brandName":"三星苹果","closeUserCode":["uc001","uc002","uc003"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8488.88","groupPrice":[{"level":"level1","boxLevelPrice":"2500"},{"level":"level2","boxLevelPrice":"3500"}],"boxPrice":[{"boxType":"box1","boxUserCode":["uc004","uc005","uc006","uc001"],"boxPriceDetail":3588.88},{"boxType":"box2","boxUserCode":["htd007","htd008","htd009","uc0010"],"boxPriceDetail":5588.88}],"boostValue":1.2}
{"index":{"_id":6}}
{"goodsName":"三星UA55RU7520JXXZ 51英寸 4K超高清","skuCode":"skuCode1","brandName":"三星","closeUserCode":["0"],"channelType":"cmccPlatform","shopCode":"sc00001","publicPrice":"8188.88","groupPrice":null,"boxPrice":null,"boostValue":1.2}
{"index":{"_id":7}}
{"goodsName":"三星UA55RU7520JXXZ 52英寸 4K超高清","skuCode":"skuCode2","brandName":"三星","closeUserCode":["0"],"channelType":"cmccPlatform","shopCode":"sc00001","publicPrice":"8288.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["htd002"],"boxPriceDetail":4288.88}],"boostValue":1.2}
{"index":{"_id":8}}
{"goodsName":"三星UA55RU7520JXXZ 52英寸 4K超高清","skuCode":"skuCode2","brandName":"三星","closeUserCode":["uc0022"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8288.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["uc0022"],"boxPriceDetail":4288.88}],"boostValue":1.2}
{"index":{"_id":9}}
{"goodsName":"三星UA55RU7520JXXZ 52英寸 4K超高清","skuCode":"skuCode2","brandName":"三星","closeUserCode":["uc0022"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8288.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["uc0022"],"boxPriceDetail":4288.88}],"boostValue":1.2}
{"index":{"_id":10}}
{"goodsName":"三星UA55RU7520JXXZ 52英寸 4K超高清","skuCode":"skuCode2","brandName":"三星","closeUserCode":["uc0022"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8288.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["uc0022"],"boxPriceDetail":4288.88}],"boostValue":1.8}

2、删除

对文档的删除操作支持以下类型

DELETE /<index>/_doc/<_id>

删除文档ID为2的数据：

DELETE /my_goods_20210423/_doc/2

另外，删除操作支持带多种条件的删除，可以使用_delete_by_query,
如下操纵，将删除店铺编码为“sc00002”的所有商品

POST /my_goods_20210423/_delete_by_query
{
  "query": {
    "match": {
      "shopCode": "sc00002"
    }
  }
}

3、修改

对文档的修改操作支持以下类型

POST /<index>/_update/<_id>

修改文档ID为1的文档信息

新增字段

POST /my_goods_20210423/_update/1
{
  "doc": {
    "shopName": "小王店铺"
  }
}

修改店铺名称为：“张三店铺”

POST /my_goods_20210423/_update/1
{
  "doc": {
    "shopName": "张三店铺"
  }
}

{
  "goodsName" : "苹果 51英寸 4K超高清",
  "skuCode" : "skuCode1",
  "brandName" : "苹果",
  "closeUserCode" : [
    "0"
  ],
  "channelType" : "cloudPlatform",
  "shopCode" : "sc00001",
  "publicPrice" : "8188.88",
  "groupPrice" : null,
  "boxPrice" : null,
  "boostValue" : 1.8,
  "shopName" : "张三店铺"
}

另外，更新操作还可以使用_update_by_query api，当店铺编码为"sc00002"时修改"publicPrice"为5888.00元
插入文档ID为2的店铺商品信息

POST /my_goods_20210423/_create/2
{"goodsName":"苹果 55英寸 3K超高清","skuCode":"skuCode2","brandName":"苹果","closeUserCode":["0"],"channelType":"cloudPlatform","shopCode":"sc00002","publicPrice":"6188.88","groupPrice":null,"boxPrice":null,"boostValue":1.0}

此时查询返回：

{
  "goodsName" : "苹果 55英寸 3K超高清",
  "skuCode" : "skuCode2",
  "brandName" : "苹果",
  "closeUserCode" : [
    "0"
  ],
  "channelType" : "cloudPlatform",
  "shopCode" : "sc00002",
  "publicPrice" : "6188.88",
  "groupPrice" : null,
  "boxPrice" : null,
  "boostValue" : 1.0
}

更新当店铺编码为"sc00002"时修改"publicPrice"为5888.00元

POST /my_goods_20210423/_update_by_query
{
  "script": {
    "source": "ctx._source.publicPrice=5888.00",
    "lang": "painless"
  },
  "query": {
    "term": {
      "shopCode": "sc00002"
    }
  }
}

再次查询结果

GET /my_goods_20210423/_source/2

{
  "shopCode" : "sc00002",
  "brandName" : "苹果",
  "closeUserCode" : [
    "0"
  ],
  "groupPrice" : null,
  "boxPrice" : null,
  "channelType" : "cloudPlatform",
  "boostValue" : 1.0,
  "publicPrice" : 5888.0,
  "goodsName" : "苹果 55英寸 3K超高清",
  "skuCode" : "skuCode2"
}

当有业务需要重建索引时需要用到_reindex api
索引的来源和目的地必须是已经存在的index,、index alias、或者data stream
你可以简单的将索引A reindex到索引B，当然也可以带条件的reindex到索引B
如下所示，将skuCode=skuCode2的商品信息reindex到索引my_goods_20210423_new中

POST _reindex
{
  "source": {
    "index": "my_goods_20210423",
    "query": {
      "match": {
        "skuCode": "skuCode2"
      }
    }
  },
  "dest": {
    "index": "my_goods_20210423_new"
  }
}

查询my_goods_20210423_new索引数据

GET my_goods_20210423_new/_search/

{
  "took" : 5,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 4,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "my_goods_20210423_new",
        "_type" : "_doc",
        "_id" : "7",
        "_score" : 1.0,
        "_source" : {
          "goodsName" : "三星UA55RU7520JXXZ 52英寸 4K超高清",
          "skuCode" : "skuCode2",
          "brandName" : "三星",
          "closeUserCode" : [
            "0"
          ],
          "channelType" : "cmccPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8288.88",
          "groupPrice" : null,
          "boxPrice" : [
            {
              "boxType" : "box1",
              "boxUserCode" : [
                "htd002"
              ],
              "boxPriceDetail" : 4288.88
            }
          ],
          "boostValue" : 1.2
        }
      },
      {
        "_index" : "my_goods_20210423_new",
        "_type" : "_doc",
        "_id" : "8",
        "_score" : 1.0,
        "_source" : {
          "goodsName" : "三星UA55RU7520JXXZ 52英寸 4K超高清",
          "skuCode" : "skuCode2",
          "brandName" : "三星",
          "closeUserCode" : [
            "uc0022"
          ],
          "channelType" : "cloudPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8288.88",
          "groupPrice" : null,
          "boxPrice" : [
            {
              "boxType" : "box1",
              "boxUserCode" : [
                "uc0022"
              ],
              "boxPriceDetail" : 4288.88
            }
          ],
          "boostValue" : 1.2
        }
      },
      {
        "_index" : "my_goods_20210423_new",
        "_type" : "_doc",
        "_id" : "9",
        "_score" : 1.0,
        "_source" : {
          "goodsName" : "三星UA55RU7520JXXZ 52英寸 4K超高清",
          "skuCode" : "skuCode2",
          "brandName" : "三星",
          "closeUserCode" : [
            "uc0022"
          ],
          "channelType" : "cloudPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8288.88",
          "groupPrice" : null,
          "boxPrice" : [
            {
              "boxType" : "box1",
              "boxUserCode" : [
                "uc0022"
              ],
              "boxPriceDetail" : 4288.88
            }
          ],
          "boostValue" : 1.2
        }
      },
      {
        "_index" : "my_goods_20210423_new",
        "_type" : "_doc",
        "_id" : "10",
        "_score" : 1.0,
        "_source" : {
          "goodsName" : "三星UA55RU7520JXXZ 52英寸 4K超高清",
          "skuCode" : "skuCode2",
          "brandName" : "三星",
          "closeUserCode" : [
            "uc0022"
          ],
          "channelType" : "cloudPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8288.88",
          "groupPrice" : null,
          "boxPrice" : [
            {
              "boxType" : "box1",
              "boxUserCode" : [
                "uc0022"
              ],
              "boxPriceDetail" : 4288.88
            }
          ],
          "boostValue" : 1.8
        }
      }
    ]
  }
}

4、查询

对文档的查询操作支持以下类型

GET <index>/_doc/<_id>

HEAD <index>/_doc/<_id>

GET <index>/_source/<_id>

HEAD <index>/_source/<_id>

查询文档ID为1的文档信息

GET /my_goods_20210423/_doc/1

查询文档ID为1的文档是否存在

只判断文档是否存在，head返回的信息更少、性能更高，满足特殊业务场景使用

HEAD /my_goods_20210423/_doc/1

200 - OK

只返回文档信息

查询时只返回_source信息

GET /my_goods_20210423/_source/1

{
  "goodsName" : "苹果 51英寸 4K超高清",
  "skuCode" : "skuCode1",
  "brandName" : "苹果",
  "closeUserCode" : [
    "0"
  ],
  "channelType" : "cloudPlatform",
  "shopCode" : "sc00001",
  "publicPrice" : "8188.88",
  "groupPrice" : null,
  "boxPrice" : null,
  "boostValue" : 1.8
}

定制化返回参数

只获取_source部分参数，类似数据库查询中的指定字段，而不是select * 返回所有字段

GET my_goods_20210423/_source/1/?_source_includes=brandName,goodsName

{
  "brandName" : "苹果",
  "goodsName" : "苹果 51英寸 4K超高清"
}

查询文档ID为1的文档是否存在

只判断文档是否存在，head返回的信息更少、性能更高，满足特殊业务场景使用

HEAD /my_goods_20210423/_doc/1

200 - OK

批量查询

ES同时支持批量查询，需要使用_mget API，查询文档ID等于1和2的文档信息

GET /my_goods_20210423/_mget
{
  "docs": [
    {
      "_id": "1"
    },
    {
      "_id": "2"
    }
  ]
}

{
  "docs" : [
    {
      "_index" : "my_goods_20210423",
      "_type" : "_doc",
      "_id" : "1",
      "_version" : 7,
      "_seq_no" : 8,
      "_primary_term" : 1,
      "found" : true,
      "_source" : {
        "goodsName" : "苹果 51英寸 4K超高清",
        "skuCode" : "skuCode1",
        "brandName" : "苹果",
        "closeUserCode" : [
          "0"
        ],
        "channelType" : "cloudPlatform",
        "shopCode" : "sc00001",
        "publicPrice" : "8188.88",
        "groupPrice" : null,
        "boxPrice" : null,
        "boostValue" : 1.8,
        "shopName" : "张三店铺"
      }
    },
    {
      "_index" : "my_goods_20210423",
      "_type" : "_doc",
      "_id" : "2",
      "found" : false
    }
  ]
}

Query DSL

查询索引包括全文本查询、组合查询、结构化查询等，主要分为query与filter查询。
2者查询是有区别的：

query查询，用于解答文档是否存在并且告知返回文档与查询条件的匹配度，返回_score评分供用户选择
filter查询，只用于返回文档是否与查询匹配，但是不会告诉你匹配度，在做聚合查询时filter经常发挥更大的作用，因为没有评分ES的处理速度就会提高，提升了整体响应时间。同时filter可以缓存查询结果，而query则不能缓存

使用场景：如果涉及到全文检索以及评分相关业务使用query，其他场景推荐使用filter查询

组合查询

image.png

boolean查询

boolean 查询包含must、filter、should、must_not
must为必须匹配并且返回评分，filter忽略评分，should相当于数据库查询中的or,must_not 为不匹配，相当于不等于
查询：店铺编码=sc00001 且渠道channelType=cloudPlatform 且publicPrice价格区间不在8288-8888之间或者品牌包含苹果

POST /my_goods_20210423/_search
{
  "query": {
    "bool": {
      "must": {
        "term":{
          "shopCode":"sc00001"
        }
      },
      "filter": {
        "term": {
          "channelType": "cloudPlatform"
        }
      },
      "must_not": [
        {
         "range": {
           "publicPrice": {
             "gte": 8288,
             "lte": 8888
           }
         }
        }
      ],
      "should": [
        {
          "term": {
            "brandName": {
              "value": "苹果"
            }
          }
        }
      ],
      "minimum_should_match" : 1
    }
  }
}

boosting 查询

boosting用于控制评分相关度相关，可以提升评分也可以降低评分

POST /my_goods_20210423/_search
{
  "query": {
    "boosting": {
      "positive": {
        "term": {
          "skuCode": {
            "value": "skuCode1"
          }
        }
      },
      "negative": {
        "term": {
          "goodsName": {
            "value": "三星"
          }
        }
      }, 
      "negative_boost": 1
    }
  }
}

此时设置的negative_boost=1，不提升也不降低，返回：

"hits" : [
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.3862942,
        "_source" : {
          "goodsName" : "苹果 51英寸 4K超高清",
          "skuCode" : "skuCode1",
          "brandName" : "苹果",
          "closeUserCode" : [
            "0"
          ],
          "channelType" : "cloudPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8188.88",
          "groupPrice" : null,
          "boxPrice" : null,
          "boostValue" : 1.8,
          "shopName" : "张三店铺"
        }
      },
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "6",
        "_score" : 1.3862942,
        "_source" : {
          "goodsName" : "三星UA55RU7520JXXZ 51英寸 4K超高清",
          "skuCode" : "skuCode1",
          "brandName" : "三星",
          "closeUserCode" : [
            "0"
          ],
          "channelType" : "cmccPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8188.88",
          "groupPrice" : null,
          "boxPrice" : null,
          "boostValue" : 1.2
        }
      }
    ]

可以看到2条文档记录评分一致："_score" : 1.3862942

当我们修改 "negative_boost": 0.2时，此时返回（省略部分无关字段）：

"hits" : [
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.3862942,
        "_source" : {
          "goodsName" : "苹果 51英寸 4K超高清",
          "skuCode" : "skuCode1",
          "brandName" : "苹果",
          "closeUserCode" : [
            "0"
          ],
          "channelType" : "cloudPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8188.88",
          "groupPrice" : null,
          "boxPrice" : null,
          "boostValue" : 1.8,
          "shopName" : "张三店铺"
        }
      },
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "6",
        "_score" : 0.27725884,
        "_source" : {
          "goodsName" : "三星UA55RU7520JXXZ 51英寸 4K超高清",
          "skuCode" : "skuCode1",
          "brandName" : "三星",
          "closeUserCode" : [
            "0"
          ],
          "channelType" : "cmccPlatform",
          "shopCode" : "sc00001",
          "publicPrice" : "8188.88",
          "groupPrice" : null,
          "boxPrice" : null,
          "boostValue" : 1.2
        }
      }
    ]

此时发现文档ID=6的评分下降到_score" : 0.27725884，因为在negative命中了查询条件，negative_boost在0到1之间时，用于降低评分，相反，大于1用于提升评分

Constant score query 查询

当查询不关心TF（词频）时，就可以使用constant score query

POST /my_goods_20210423/_search
{
  "query": {
    "constant_score": {
      "filter": {
        "term": {
          "goodsName": "苹果"
        }
      },
      "boost": 1.2
    }
  }
}

返回（省略部分无关字段）：

{
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "3",
        "_score" : 1.2,
        "_source" : {
          "goodsName" : "苹果UA55RU7520JXXZ 53英寸 4K高清"
        }
      },
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "4",
        "_score" : 1.2,
        "_source" : {
          "goodsName" : "山东苹果UA55RU7520JXXZ 苹果54英寸 5K超高清"
        }
      }
}

可以看到，文档ID=3的评分和文档ID=4的评分一样，但是ID=4的匹配相关度更高，这是由于我们忽略了词频对打分的影响。

Disjunction max query 查询

Disjunction 查询也被理解为分离最大化查询，指的是：将任何与任一查询匹配的文档作为结果返回，但只将最佳匹配的评分作为查询的评分结果返回，例如查询商品名称和品牌名称中包含“苹果”的信息，当品牌的评分高于商品名称时，则返回品牌的评分做为总评分（忽略tie_breaker缓冲）

GET /my_goods_20210423/_search
{
  "query": {
    "dis_max": {
      "tie_breaker": 0.7,
      "boost": 1.2,
      "queries": [
        {
          "term": {
            "goodsName": {
              "value": "苹果"
            }
          }
        },
        {
          "term": {
            "brandName": {
              "value": "苹果"
            }
          }
        }
        ]
    }
  }
}

返回结果（忽略无关字段）：

"max_score" : 3.0150018,
    "hits" : [
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 3.0150018,
        "_source" : {
          "goodsName" : "苹果 51英寸 4K超高清",
          "brandName" : "苹果"
        }
      },
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "5",
        "_score" : 1.3465583,
        "_source" : {
          "goodsName" : "苹果UA55R苹果U7苹果520JXXZ 55英寸 5K超高清",
          "brandName" : "三星苹果"
        }
      },
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "4",
        "_score" : 1.2337791,
        "_source" : {
          "goodsName" : "山东苹果UA55RU7520JXXZ 苹果54英寸 5K超高清",
          "brandName" : "山东苹果"
        }
      },

分析：

id=1的记录，由于品牌只包含“苹果”2字，ES认为这种匹配度更高，所以此条记录评分排在第一位
id=5的记录，由于品牌中和ID=4的记录都包含苹果且字数一样，此时就要看goodsName包含苹果的词频数量了，ID=5的品牌中，“苹果”出现了3次，而ID=4的值出现了2次，所以评分没有ID=5的高，符合我们的预期结果。
tie_breaker字段做什么用呢？它是起到了缓冲的作用（取值范围：0到1之间），Disjunction查询会将匹配度最高的字段得分做为整个文档的得分返回，这种情况其他字段就不起作用了，难免有点走极端，此时就需要tie_breaker来做缓存，提升其他字段的影响力，最终的结果：brandName评分+goodsName评分*tie_breaker。作为总评分返回

Function score query 查询

Function score 允许你控制查询评分，是用来控制评分过程的终极武器。最高效的用法是用过滤器对结果的子集应用不同的函数，同时运用了filter的缓存并且达到了控制评分的过程。
我们想让山东的苹果搜索出现美国苹果之前，查询商品名称包含“苹果”，当品牌中包含“美国”时，权重设置为2，当出现“山东”时，权重设置为40

GET /my_goods_20210423/_search
{
  "query": {
    "function_score": {
      "query": {
        "term": {
          "goodsName": {
            "value": "苹果"
          }
        }
      },
      "boost": 2, 
      "functions": [
        {
          "filter": {
            "match":{
              "brandName":"美国"
            }
          },
          "random_score": {
            
          },
          "weight": 2
        },
        {
          "filter": {
            "match":{
              "brandName":"山东"
            }
          },
          "weight": 40
        }
      ],
      "max_boost": 60,
      "score_mode": "max",
      "boost_mode": "multiply",
      "min_score": 2
    }
  }
}

返回主要信息：

    "max_score" : 2.2442641,
    "hits" : [
     {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "4",
        "_score" : 2.0562985,
        "_source" : {
          "goodsName" : "山东苹果UA55RU7520JXXZ 苹果54英寸 5K超高清",
          "brandName" : "山东苹果"
        }
      },
      {
        "_index" : "my_goods_20210423",
        "_type" : "_doc",
        "_id" : "3",
        "_score" : 1.7582327,
        "_source" : {
          "goodsName" : "苹果UA55RU7520JXXZ 53英寸 4K高清",
          "brandName" : "美国苹果",
        }
      }
    ]

解释几个参数：

score_mode
multiply:默认，分数相乘
sum：分数求和
avg：平均分数
first：第一个 function的分数
max：使用评分最大的分数
min：使用评分最小的分数

avg举例，如果2个函数返回的分数为1和2，并且它们的权重分别为3和4，则他们的评分为：(13+24)/(3+4)
其他详解请参考官方score-functions详解

全文检索

全文检索.png

match 查询

match 查询是一种标准的查询，示例如下：

GET /my_goods_20210423/_search
{
  "query": {
    "match": {
      "goodsName": "苹果 高清 英寸"
    }
  }
}

match查询是一种boolean类型的查询，可以使用"operator"来控制boolean 字句，operator包含 and 和 or(默认为 or)

GET /my_goods_20210423/_search
{
  "query": {
    "match": {
      "goodsName": {
        "query": "苹果 高清 英寸",
        "operator": "and"
      }
    }
  }
}

返回结果：

{
  "took" : 1,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 0,
      "relation" : "eq"
    },
    "max_score" : null,
    "hits" : [ ]
  }
}

命中为0，因为没有标题中包含“苹果高清英寸”词组的商品信息

match boolean prefix query

添加2条商品名称是因为的测试数据，方便测试

POST my_goods_20210423/_bulk
{"index":{"_id":11}}
{"goodsName":"apple goods test","skuCode":"skuCode3","brandName":"美国苹果","closeUserCode":["0"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8388.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["htd003","uc004"],"boxPriceDetail":4388.88},{"boxType":"box2","boxUserCode":["uc005","uc0010"],"boxPriceDetail":5388.88}],"boostValue":1.2}
{"index":{"_id":12}}
{"goodsName":"apple goods online","skuCode":"skuCode3","brandName":"美国苹果","closeUserCode":["0"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8388.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["htd003","uc004"],"boxPriceDetail":4388.88},{"boxType":"box2","boxUserCode":["uc005","uc0010"],"boxPriceDetail":5388.88}],"boostValue":1.2}

GET /my_goods_20210423/_search
{
  "query": {
    "match_bool_prefix": {
      "goodsName": "apple goods t"
    }
  }
}

2条刚添加的商品都被查询到了，match_bool_prefix原理就相当于把词组分开后的boolean查询，转换后类似如下查询：

GET /my_goods_20210423/_search
{
  "query": {
    "bool" : {
      "should": [
        { "term": { "goodsName": "apple" }},
        { "term": { "goodsName": "goods" }},
        { "prefix": { "goodsName": "t"}}
      ]
    }
  }
}

match prefix query

用于匹配索引中是否存在所输入的查询条件数据

GET /my_goods_20210423/_search
{
  "query": {
    "match_phrase": {
      "goodsName": "apple"
    }
  }
}

比较match_phrase与match区别，match_phrase会将查询条件的中的信息看做一个整体，不做分词去查询，当然你也可以指定分词类型，而match会将查询中的条件做分词处理后，再去做查询

#查询不到任何数据，因为不存在'goods t'的词组
GET /my_goods_20210423/_search
{
  "query": {
    "match_phrase": {
      "goodsName": "goods t"
    }
  }
}
#能查询到数据，因为文档中包含goods和t的词组
GET /my_goods_20210423/_search
{
  "query": {
    "match": {
      "goodsName": "goods t"
    }
  }
}

match phrase prefix query

返回文档包含给定查询条件的文档，文档中必须包含给定条件的内容且是按照顺序的，如"apple goods t" ，商品名称包含"apple goods test"的数据将被查询到返回。
新增一条测试数据

POST my_goods_20210423/_bulk
{"index":{"_id":13}}
{"goodsName":"apple and goods product ","skuCode":"skuCode3","brandName":"美国苹果","closeUserCode":["0"],"channelType":"cloudPlatform","shopCode":"sc00001","publicPrice":"8388.88","groupPrice":null,"boxPrice":[{"boxType":"box1","boxUserCode":["htd003","uc004"],"boxPriceDetail":4388.88},{"boxType":"box2","boxUserCode":["uc005","uc0010"],"boxPriceDetail":5388.88}],"boostValue":1.2}

#只返回goodsName : apple goods test的数据
GET /my_goods_20210423/_search
{
  "query": {
    "match_phrase_prefix": {
      "goodsName": "apple goods t"
    }
  }
}

总结比较match这四种查询

match比较.png

Multi-match

多字段匹配，可以在多个字段中匹配查询相关信息，通过type参数可以调整结果集

#查询商品名称和品牌名称中包含苹果的文档信息
POST /my_goods_20210423/_search
{
  "query": {
    "multi_match": {
      "query": "苹果",
      "type": "best_fields", 
      "fields": ["goodsName","brandName"],
      "tie_breaker": 0.3
    }
  }
}

type参数类型详解：

best_fields ：默认，匹配fields，将评分最高的分数做为整个查询的分数返回
most_fields：查询匹配的文档，并且返回各个字段的分数之和的平均值
cross_fields：跨字段匹配，匹配多个字段中是否包含查询词组
phrase：以match_phrase方式运行查询，并返回最佳匹配的评分做为总评分
phrase_prefix：以match_phrase_prefix方式运行查询，并返回最佳匹配的评分做为总评分
bool_prefix：在每个字段上运行match_bool_prefix查询并组合每个字段的评分，详情参考bool_prefix
以cross_fields为例进行实战讲解

#插入测试数据
PUT my_shop
{
  "settings": {
    "number_of_shards": 1,
    "number_of_replicas": 1
  },
  "mappings": {
    "properties": {
      "firstName":{
        "type":"text"
      },
      "lastName":{
        "type":"text"
      }
    }
  }
}
POST my_shop/_bulk
{"index":{"_id":1}}
{"first_name":"Will","last_name":"Smith","age":25}
{"index":{"_id":2}}
{"first_name":"Smith","last_name":"hello","age":21}
{"index":{"_id":3}}
{"first_name":"Will","last_name":"hello","age":20}

#查询姓名为Will Smith的信息
GET /my_shop/_search
{
  "query": {
    "multi_match" : {
      "query":      "Will Smith",
      "type":       "cross_fields",
      "fields":     [ "first_name^2", "last_name" ],
      "operator":   "and"
    }
  }
}
#返回
"max_score" : 1.9208363,
    "hits" : [
      {
        "_index" : "my_shop",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.9208363,
        "_source" : {
          "first_name" : "Will",
          "last_name" : "Smith",
          "age" : 25
        }
      }
    ]

另外，first_name提升了权重，默认为1

Term-level查询

可以使用term-level 查询结构化数据，结构化数据如日期范围、IP地址、价格等,下面分别演示在业务场景中的实际使用

exists查询
返回包含字段索引值的文档

#返回包含goodsName字段的索引文档
GET /my_goods_20210423/_search
{
  "query": {
    "exists": {
      "field": "goodsName"
    }
  }
}

fuzzy查询
返回包含与搜索字词相似的字词的文档,可以用于查询纠错功能

#以官网例子举例说明
POST /my_index/_bulk
{ "index": { "_id": 1 }}
{ "text": "Surprise me!"}
{ "index": { "_id": 2 }}
{ "text": "That was surprising."}
{ "index": { "_id": 3 }}
{ "text": "I wasn't surprised."}

GET /my_index/_search
{
  "query": {
    "fuzzy": {
      "text": {
        "value": "surprize",
        "prefix_length": 1
      }
    }
  }
}
#发挥
"hits" : [
      {
        "_index" : "my_index",
        "_type" : "my_type",
        "_id" : "1",
        "_score" : 0.9559981,
        "_source" : {
          "text" : "Surprise me!"
        }
      },
      {
        "_index" : "my_index",
        "_type" : "my_type",
        "_id" : "3",
        "_score" : 0.69983494,
        "_source" : {
          "text" : "I wasn't surprised."
        }
      }

默认如果不设置，prefix_length就是2

surprising 错误3个位置，不能纠错
surprize 拼写错误，s->z,错误在一个位置，在2个位置的纠错范围之内
为提高性能，可以设置max_expansions，将限制产生模糊文档的个数，
另外，prefix_length不宜设置过大，也将影响查询性能，同时错误过多也将导致查询结果不是用户期望的。

ids查询
范围文档包含ID的文档信息

GET /my_goods_20210423/_search
{
  "query": {
    "ids" : {
      "values" : ["1", "4", "5"]
    }
  }
}

prefix查询
返回在提供的字段中包含特定前缀的文档

GET /my_shop_test/_search
{
  "query": {
    "prefix": {
      "shopName": {
        "value": "bo"
      }
    }
  }
}
#返回
"hits" : [
      {
        "_index" : "my_shop_test",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "shopName" : "box",
          "shopCode" : "Smith"
        }
      },
      {
        "_index" : "my_shop_test",
        "_type" : "_doc",
        "_id" : "4",
        "_score" : 1.0,
        "_source" : {
          "shopName" : "booex",
          "shopCode" : "act"
        }
      }
    ]

range查询
rand查询类似数据库中的大于、小于范围查询

GET my_goods_20210423/_search
{
  "query": {
    "range": {
      "publicPrice": {
        "gte": 2000,
        "lte": 8488
      }
    }
  }
}

gt :大于
gte：大于等于
lt:小于
lte：小于等于

regexp查询
正则表达式查询，查询店铺编码以's'开头，中间包括任何字符以及长度并且以'1'结尾的数据

GET my_goods_20210423/_search
{
  "query": {
    "regexp": {
      "shopCode": {
        "value": "s.*1",
        "flags": "ALL",
        "case_insensitive": true,
        "max_determinized_states": 10000,
        "rewrite": "constant_score"
      }
    }
  }
}

term查询

#返回确切的文档内容，避免对text字段类型使用term
GET my_goods_20210423/_search
{
  "query": {
    "term": {
      "brandName": {
        "value": "三星",
        "boost": 1.0
      }
    }
  }
}

terms查询
terms返回一个或多个包含精确查询条件的文档信息

GET /my_goods_20210423/_search
{
  "query": {
    "terms": {
      "brandName": [ "美国", "三星" ],
      "boost": 1.0
    }
  }
}

terms_set查询
返回最小精确匹配成功的文档信息，terms_set类似terms 查询，只不过terms_se多定义了返回最小匹配的数量

#新定义商品信息
PUT /my_goods_info
{
  "mappings": {
    "properties": {
      "goodsName": {
        "type": "keyword"
      },
      "sale_property": {
        "type": "keyword"
      },
      "required_matches": {
        "type": "long"
      }
    }
  }
}

#添加3条商品测试数据
#销售属性 白色、64G、标品
PUT /my_goods_info/_doc/1?refresh
{
  "name": "apple",
  "sale_property": [ "white", "64","standard" ],
  "required_matches": 2
}
#黑色、32G、非标品
PUT /my_goods_info/_doc/2?refresh
{
  "name": "apple",
  "sale_property": [ "black", "32","no standard" ],
  "required_matches": 2
}
#黑色、64 非标品
PUT /my_goods_info/_doc/3?refresh
{
  "name": "apple",
  "sale_property": [ "black", "64","no standard" ],
  "required_matches": 2
}
#查询
GET /my_goods_info/_search
{
  "query": {
    "terms_set": {
      "sale_property": {
        "terms": [ "white", "64"],
        "minimum_should_match_field": "required_matches"
      }
    }
  }
}
#返回
"hits" : [
      {
        "_index" : "my_goods_info",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.1149836,
        "_source" : {
          "name" : "apple",
          "sale_property" : [
            "white",
            "64",
            "standard"
          ],
          "required_matches" : 2
        }
      }
    ]

wildcard查询
返回包含与通配符模式匹配的术语的文档

#返回
GET /my_goods_20210423/_search
{
  "query": {
    "wildcard": {
      "shopCode": {
        "value": "sc*1",
        "boost": 1.0,
        "rewrite": "constant_score"
      }
    }
  }
}

最后编辑于：2021.04.27 20:52:25

人面猴
序言：七十年代末，一起剥皮案震惊了整个滨河市，随后出现的几起案子，更是在滨河造成了极大的恐慌，老刑警刘岩，带你破解...
沈念sama阅读 158,560评论 4赞 361
死咒
序言：滨河连续发生了三起死亡事件，死亡现场离奇诡异，居然都是意外死亡，警方通过查阅死者的电脑和手机，发现死者居然都...
沈念sama阅读 67,104评论 1赞 291
救了他两次的神仙让他今天三更去死
文/潘晓璐我一进店门，熙熙楼的掌柜王于贵愁眉苦脸地迎上来，“玉大人，你说我怎么就摊上这事。” “怎么了？”我有些...
开封第一讲书人阅读 108,297评论 0赞 243
道士缉凶录：失踪的卖姜人
文/不坏的土叔我叫张陵，是天一观的道长。经常有香客问我，道长，这世上最难降的妖魔是什么？我笑而不...
开封第一讲书人阅读 43,869评论 0赞 204
港岛之恋（遗憾婚礼）
正文为了忘掉前任，我火速办了婚礼，结果婚礼上，老公的妹妹穿的比我还像新娘。我一直安慰自己，他们只是感情好，可当我...
茶点故事阅读 52,275评论 3赞 287
恶毒庶女顶嫁案：这布局不是一般人想出来的
文/花漫我一把揭开白布。她就那样静静地躺着，像睡着了一般。火红的嫁衣衬着肌肤如雪。梳的纹丝不乱的头发上，一...
开封第一讲书人阅读 40,563评论 1赞 216
城市分裂传说
那天，我揣着相机与录音，去河边找鬼。笑死，一个胖子当着我的面吹牛，可吹牛的内容都是我干的。我是一名探鬼主播，决...
沈念sama阅读 31,833评论 2赞 312
双鸳鸯连环套：你想象不到人心有多黑
文/苍兰香墨我猛地睁开眼，长吁一口气：“原来是场噩梦啊……” “哼！你这毒妇竟也来了？” 一声冷哼从身侧响起，我...
开封第一讲书人阅读 30,543评论 0赞 197
万荣杀人案实录
序言：老挝万荣一对情侣失踪，失踪者是张志新（化名）和其女友刘颖，没想到半个月后，有当地人在树林里发现了一具尸体，经...
沈念sama阅读 34,245评论 1赞 241
护林员之死
正文独居荒郊野岭守林人离奇死亡，尸身上长有42处带血的脓包…… 初始之章·张勋以下内容为张勋视角年9月15日...
茶点故事阅读 30,512评论 2赞 244
白月光启示录
正文我和宋清朗相恋三年，在试婚纱的时候发现自己被绿了。大学时的朋友给我发了我未婚夫和他白月光在一起吃饭的照片。...
茶点故事阅读 32,011评论 1赞 258
活死人
序言：一个原本活蹦乱跳的男人离奇死亡，死状恐怖，灵堂内的尸体忽然破棺而出，到底是诈尸还是另有隐情，我是刑警宁泽，带...
沈念sama阅读 28,359评论 2赞 253
日本核电站爆炸内幕
正文年R本政府宣布，位于F岛的核电站，受9级特大地震影响，放射性物质发生泄漏。R本人自食恶果不足惜，却给世界环境...
茶点故事阅读 33,006评论 3赞 235
男人毒药：我在死后第九天来索命
文/蒙蒙一、第九天我趴在偏房一处隐蔽的房顶上张望。院中可真热闹，春花似锦、人声如沸。这庄子的主人今日做“春日...
开封第一讲书人阅读 26,062评论 0赞 8
一桩弑父案，背后竟有这般阴谋
文/苍兰香墨我抬头看了看天上的太阳。三九已至，却和暖如春，着一层夹袄步出监牢的瞬间，已是汗流浃背。一阵脚步声响...
开封第一讲书人阅读 26,825评论 0赞 194
情欲美人皮
我被黑心中介骗来泰国打工，没想到刚下飞机就差点儿被人妖公主榨干…… 1. 我叫王不留，地道东北人。一个月前我还...
沈念sama阅读 35,590评论 2赞 273
代替公主和亲
正文我出身青楼，却偏偏与公主长得像，于是被迫代替她去往敌国和亲。传闻我的和亲对象是个残疾皇子，可洞房花烛夜当晚...
茶点故事阅读 35,501评论 2赞 268

Elasticsearch 搜索基本操作

业务背景

定义mapping

商品字段描述如下：

定义商品mapping

操作文档

1、新增

2、删除

3、修改

修改文档ID为1的文档信息

4、查询

查询文档ID为1的文档信息

查询文档ID为1的文档是否存在

只返回文档信息

定制化返回参数

查询文档ID为1的文档是否存在

批量查询

Query DSL

使用场景：如果涉及到全文检索以及评分相关业务使用query，其他场景推荐使用filter查询

组合查询

boolean查询

boosting 查询

Constant score query 查询

Disjunction max query 查询

Function score query 查询

全文检索

match 查询

match boolean prefix query

match prefix query

match phrase prefix query

总结比较match这四种查询

Multi-match

Term-level查询

推荐阅读更多精彩内容