Loading

(一)《SQL进阶教程》学习记录--CASE

背景:最近用到统计之类的复杂Sql比较多,有种“提笔忘字”的感觉,看书练习,举一反三,巩固加强。💪

(一) 《SQL进阶教程》学习记录--CASE
(二) 《SQL进阶教程》学习记录--GROUP BY、PARTITION BY

1、语法

两种写法:简单 CASE 表达式(simple case expression)、搜索 CASE 表达式(searched case expression)

-- 简单 CASE 表达式
CASE sex
WHEN '1' THEN '男'
WHEN '2' THEN '女'
ELSE '其他' END

-- 搜索 CASE 表达式
CASE WHEN sex = '1' THEN '男'
WHEN sex = '2' THEN '女'
ELSE '其他' END

-- 可以用in
CASE WHEN address IN ('石家庄', '邯郸') THEN '河北'
WHEN address = '郑州' THEN '河南'
ELSE '其他' END

例1:统计各省人数,people表结构如下

SELECT
CASE		
  WHEN area IN ( '石家庄', '邯郸', '秦皇岛' ) THEN '河北' 
  WHEN area IN ( '郑州', '安阳' ) THEN '河南' 
  ELSE '其他' END as province,
  SUM ( population ) as population
  FROM people 
  GROUP BY province

看似没毛病,其实是有坑的,不同数据库是有区别的,这句在 PostgreSQL 和 MySQL 可以顺利执行,在 Oracle、 DB2、SQL Server 会报错province不存在,需要修改一下

SELECT province,SUM ( population ) as population from 
(SELECT
   CASE		
    WHEN area IN ( '石家庄', '邯郸', '秦皇岛' ) THEN '河北' 
    WHEN area IN ( '郑州', '安阳' ) THEN '河南' 
    ELSE '其他' END as province , population 
    FROM people ) peoplesum 
    GROUP BY province

其实,对于使用过多个库的开发人员,看错误也能八九不离十的写出来,重点是下面这种通用SQL,以前还真没Get过,就是把条件在写一遍,可能是因为会变长吧,哈哈哈哈~~~嗝

SELECT
CASE		
  WHEN area IN ( '石家庄', '邯郸', '秦皇岛' ) THEN '河北' 
  WHEN area IN ( '郑州', '安阳' ) THEN '河南' 
  ELSE '其他' END as province,
  SUM ( population ) as population
  FROM people 
  GROUP BY
  CASE		
    WHEN area IN ( '石家庄', '邯郸', '秦皇岛' ) THEN '河北' 
    WHEN area IN ( '郑州', '安阳' ) THEN '河南' 
    ELSE'其他' END

以上三种写法,结果一致

2、行转列

高频“考点”,study表结构如下

SELECT
 username,
 SUM ( CASE subject WHEN '语文' THEN fraction ELSE 0 END ) AS yuwen,
 SUM ( CASE subject WHEN '数学' THEN fraction ELSE 0 END ) AS shuxue,
 SUM ( CASE subject WHEN '英语' THEN fraction ELSE 0 END ) AS yingyu 
FROM
 study 
GROUP BY
 username

3、条件更新

公司受疫情影响,全员降薪,挣得多降的多。月薪>=10w,减20%,10w>月薪>=5,减10%,5w>月薪>=1w,减5%,月薪<1w,不变(毕竟生活太难了🤣),如表:万万不可分次执行,因为有的人工资降完又符合了更低档的要求,再降就没法活了。。。当然你可以从低到高执行三次,更简单的如下:

update salaries set salary = CASE WHEN salary >= 100000 THEN salary*0.8
				  WHEN salary >= 50000 AND salary < 100000 THEN salary*0.9
				  WHEN salary >= 10000 AND salary < 50000 THEN salary*0.95
				  ELSE salary END;

4、嵌套CASE

学以致用,真实需求搞一把。按月分别统计连续两个月高报警、低报警时长综合。简化如表:

SELECT 
  SUM(CASE to_char(create_time,'yyyy-MM') WHEN '2021-08' THEN (CASE alarm_level WHEN '高报' then time_span else 0 end) ELSE 0 END  ) as 本月高报,
  SUM(CASE to_char(create_time,'yyyy-MM') WHEN '2021-08' THEN (CASE alarm_level WHEN '低报' then time_span else 0 end) ELSE 0 END  ) as 本月低报,
  SUM(CASE to_char(create_time,'yyyy-MM') WHEN '2021-07' THEN (CASE alarm_level WHEN '高报' then time_span else 0 end) ELSE 0 END  ) as 上月高报,
  SUM(CASE to_char(create_time,'yyyy-MM') WHEN '2021-07' THEN (CASE alarm_level WHEN '低报' then time_span else 0 end) ELSE 0 END  ) as 上月低报
FROM alarm_info

posted @ 2021-11-07 15:06  冬先生  阅读(540)  评论(0编辑  收藏  举报