Elasticsearch(六)过滤器

• Bool filter

  有must、should、must_not三种逻辑操作;其中当仅存在should时则必须至少满足一个条件.

  GET /books/book/_search
  {
    "query": {
      "bool": {
        "must": [
          {
            "match": {
              "name": "python"
            }
          }
        ],
        "should": [
          {
            "match": {
              "title": "effective"
            }
          }
        ]
      }
    }
  }

• Exists filter

  exists filter过滤搜索结果,使其必须存在某个指定的字段.

  # exists filter
  GET /books/book/_search
  {
  "query": {
    "exists": {
      "field": "name"
    }
  }
  }

Null值的讨论:

假如索引test中存在类型为test的如下文档:

{"user": ""} # 1
{"user": []} # 2
{"user": null} # 3
{"user": [null]} # 4
{"user": "-"} # 5
{"user": "foo"} # 6
{"person": "bar"} # 7

对于如下的查询,将会返回1、5、6三条数据.

默认情况下:2、3、4、7四种情况都被认为是null值而被过滤,最终不可被搜索.

GET /test/test/_search
{
  "query": {
    "exists": {
      "field": "user"
    }
  }
}

可以在创建索引前,通过设置mapping中null_value属性的值,从而让3、4这两种情况文档可被搜索.

• Type filter

  返回指定type的文档.

  GET /books/_search
  {
    "query": {
      "type": {
        "value": "book"
      }
    }
  }

• Match_all filter

  选中全部数据,相当于SQL语句中的select * from book.

  GET /books/_search
  {
    "query": {
      "type": {
        "value": "book"
      }
    }
  }

• Query filter

  由AND、OR两种逻辑

  GET /books/book/_search
  {
    "query": {
      "query_string": {
        "default_field": "name",
        "query": "python OR java"  # "python AND java"
      }
    }
  }

• 结果排序

  默认按照相关度得分_score倒排,还可以通过sort针对特定字段正排(asc)和倒排(desc);另外还可以在sort中指定_last将无值的结果放在检索集的最后.

  GET /users/user/_search
  {
    "query": {
      "match": {
        "name": "bourne"
      }
    },
    "sort": [
      {
        "created_at": {
          "order": "desc",
          "missing": "_last"
        }
      }
    ]
  }