问题场景
最近有个朋友去面试,问了我一道面试题。题目如下,在形如下面的数据库表score中,找出每门成绩(grade)都大于等于80分的学生姓名。
----------------------------------------
name
| course | grade
----------------------------------------
zhangsan | Java | 70
----------------------------------------
zhangsan | C++ | 80
----------------------------------------
lisi | Java | 90
----------------------------------------
lisi | C++ | 60
----------------------------------------
wangwu | Java | 85
----------------------------------------
wangwu | C++ | 95
----------------------------------------
期望结果
----------------------------------------
name
----------------------------------------
wangwu
----------------------------------------
本文以MySQL数据库为例,以三种方案一步一步实现面试题要求。
方案一
1、寻求行列转换,也称矩阵转置,将多列数据转为一行,这一步最关键,实现SQL语句如下:
下面的sql是以score为主表来构建多列,构造的从表为每门课程的数据表,通过主表的 name 字段来关联从表的 name 字段。
select s.`name` as name, (select grade from score where name = s.`name` and course = ‘Java‘)as ‘Java‘, (select grade from score where name = s.`name` and course = ‘C++‘)as ‘C++‘ from score s
运行结果截图:
2、对于1的执行结果,需要过滤掉重复的行,只要通过 distinct 关键字就可以了,实现SQL语句如下:
select distinct s.`name` as name,#此处加distinct来过滤相同的行 (select grade from score where name = s.`name` and course = ‘Java‘)as ‘Java‘, (select grade from score where name = s.`name` and course = ‘C++‘)as ‘C++‘ from score s
运行结果截图:
3、最后通过构造一个子查询,即是把上面2的查询结果作为一个表来查询,进行 where 行级过滤就可以了,实现SQL语句如下:
select * from ( select distinct s.`name` as name,#此处加distinct来过滤相同的行 (select grade from score where name = s.`name` and course = ‘Java‘)as G1, (select grade from score where name = s.`name` and course = ‘C++‘)as G2 from score s ) score where G1>=80 and G2>=80
运行结果截图:
问题:这里有一个问题指出下,如果写成如下的sql语句,把Java和C++作为列名的话(还有C#),查询结果为NULL,这个问题后续会详解,请见运行结果截图:
select * from ( select distinct s.`name` as name,#此处加distinct来过滤相同的行 (select grade from score where name = s.`name` and course = ‘Java‘)as ‘Java‘, (select grade from score where name = s.`name` and course = ‘C++‘)as ‘C++‘ from score s ) score where ‘Java‘>=80 and ‘C++‘>=80
运行结果截图:
方案二
1、通过group和聚合函数sum的结合使用,通过group by name分组,利用sum假模假样计算出每门课程的成绩,sum的时候利用case判断课程类别,就得到了以行显示的每个学生的每门课程成绩,这一步最关键,实现SQL语句如下:
select name, sum(case when course=‘Java‘ then grade end) as ‘Java‘, sum(case when course=‘C++‘ then grade end) as ‘C++‘ from score group by name
运行结果截图:
2、再通过构造一个子查询,即是把上面1的查询结果作为一个表来查询,进行 where 行级过滤就可以了,实现SQL语句如下:
select * from ( select name, sum(case when course=‘Java‘ then grade end) as G1, sum(case when course=‘C++‘ then grade end) as G2 from score group by name ) score where G1>=80 and G2>=80
运行结果截图:
方案三
1、先找出有任意课程<80分的学生,实现SQL语句如下:
select name from score where grade<80
运行结果截图:
2、distinct出所有学生列表(不重复),实现SQL语句如下:
select distinct name from score
运行结果截图:
3、通过构造子查询从查询2的结果排除出去查询1的结果,这一步骤有的数据库是有集合函数,比如SQL Server的Except,这儿我们用not exists进行行级过滤,实现SQL语句如下:
select * from ( select distinct name from score ) score1 where not exists ( select * from ( select distinct name from score where grade<80 ) score2 where score1.name=score2.name )
运行结果截图:
总结: