elasticsearch搜索映射Mysql

時間 2019-12-05

原文原文鏈接

這裏總結一下es徹底映射mysql的搜索語句，但願可以幫到學習es的同窗們。mysql

實例

這裏咱們假設mysql表有7個字段: birthday, name, sex, height, address, province, city. sql

實例1: mysql and + or

sql:緩存

select * from table where birthday = '2018-05-17' and name = 'xx' 
 and (sex = 1 or height = 183 or address like 'chengdu%')

es的搜索語句有兩種方式學習

方式1: must + shouldspa

{
  "query": {
    "bool": {
      "must": [
        {
          "term": {
            "birthday": "2018-05-17"
          }
        },
        {
          "term": {
            "name": "xx"
          }
        },
        {
          "bool": {
            "should": [
              {
                "term": {
                  "sex": 1
                }
              },
              {
                "term": {
                  "height": 1
                }
              },
              {
                "wildcard": {
                  "address": {
                    "wildcard": "chengdu*"
                  }
                }
              }
            ]
          }
        }
      ]
    }
  }
}

方式2: 用filter + shouldcode

{
  "query": {
    "bool": {
      "filter": [
        {
          "term": {
            "birthday": "2018-05-17"
          }
        },
        {
          "term": {
            "name": "xx"
          }
        }
      ],
      "should": [
        {
          "term": {
            "sex": 1
          }
        },
        {
          "term": {
            "height": 1
          }
        },
        {
          "wildcard": {
            "address": {
              "wildcard": "chengdu*"
            }
          }
        }
      ]
    }
  }
}

實例2: mysql 子查詢 + or

sql:ci

select * from 
 (select * from table where birthday > '2018-05-17' and birthday < '2018-05-17' and name = 'xx') 
where sex = 1 or height = 183

es的搜索語句是:文檔

{
  "query": {
    "bool": {
      "filter": [
        {
          "range": {
            "birthday": {
              "from": "2018-05-17",
              "include_lower": true,
              "include_upper": true,
              "to": null
            }
          }
        },
        {
          "range": {
            "birthday": {
              "from": null,
              "include_lower": true,
              "include_upper": true,
              "to": "2018-05-24"
            }
          }
        },
        {
          "match": {
            "name ": {
              "query": "xx"  // 這裏不該該對應mysql的=, term應該是=. match用到的是es裏面的分詞,但mysql找不到與之匹配的.
            }
          }
        }
      ],
      "should": [
        {
          "term": {
            "sex": 1
          }
        },
        {
          "term": {
            "height ": 183
          }
        }
      ]
    }
  }
}

實例3: and 與 or 嵌套

sql:it

select * from table where birthday = '2018-05-17' 
 and ( (sex = 1 and height = 183) or (province = 'chengdu' or city = 'chengdu'))

es搜索語句:table

{
  "query": {
    "bool": {
      "filter": [
        {
          "term": {
            "birthday": "2018-05-17"
          }
        }
      ],
      "must": {
        "bool": {
          "should": [
            {
              "bool": {
                "must": [
                  {
                    "term": {
                      "sex": 1
                    }
                  },
                  {
                    "term": {
                      "height": 183
                    }
                  }
                ]
              }
            },
            {
              "multi_match": {  // 能夠用should代替
                "fields": [
                  "province",
                  "city"
                ],
                "query": "chengdu"
              }
            }
          ]
        }
      }
    }
  }
}

這三種狀況，應該能應對絕大多數的條件查詢場景。

值得注意的filter和query, 因爲filter不包括評分查詢, 因此更快,更穩定，結果容易緩存，但結果可能不許確。 query是評分查詢（scoring queries）不單單要找出匹配的文檔，還要計算每一個匹配文檔的相關性，計算相關性使得它們比不評分查詢費力的多。同時，查詢結果並不緩存。