分组取前N记录 [转]

原文链接：http://blog.csdn.net/acmain_chm/article/details/4126306

经常看到问题，如何取出每组的前N条记录。方便大家参考于是便把常见的几种解法列出于下。

问题：有表 如下，要求取出各班前两名（允许并列第二）
Table1
+----+------+------+-----+
| id |SName |ClsNo |Score|
+----+------+------+-----+
|  1 |AAAA  |  C1  | 67  |
|  2 |BBBB  |  C1  | 55  |
|  3 |CCCC  |  C1  | 67  |
|  4 |DDDD  |  C1  | 65  |
|  5 |EEEE  |  C1  | 95  |
|  6 |FFFF  |  C2  | 57  |
|  7 |GGGG  |  C2  | 87  |
|  8 |HHHH  |  C2  | 74  |
|  9 |IIII  |  C2  | 52  |
| 10 |JJJJ  |  C2  | 81  |
| 11 |KKKK  |  C2  | 67  |
| 12 |LLLL  |  C2  | 66  |
| 13 |MMMM  |  C2  | 63  |
| 14 |NNNN  |  C3  | 99  |
| 15 |OOOO  |  C3  | 50  |
| 16 |PPPP  |  C3  | 59  |
| 17 |QQQQ  |  C3  | 66  |
| 18 |RRRR  |  C3  | 76  |
| 19 |SSSS  |  C3  | 50  |
| 20 |TTTT  |  C3  | 50  |
| 21 |UUUU  |  C3  | 64  |
| 22 |VVVV  |  C3  | 74  |
+----+------+------+-----+
结果如下
+----+------+------+-----+
| id |SName |ClsNo |Score|
+----+------+------+-----+
|  5 |EEEE  |  C1  | 95  |
|  1 |AAAA  |  C1  | 67  |
|  3 |CCCC  |  C1  | 67  |
|  7 |GGGG  |  C2  | 87  |
| 10 |JJJJ  |  C2  | 81  |
| 14 |NNNN  |  C3  | 99  |
| 18 |RRRR  |  C3  | 76  |
+----+------+------+-----+

方法一：
select a.id,a.SName,a.ClsNo,a.Score
from Table1 a left join Table1 b on a.ClsNo=b.ClsNo and a.Score<b.Score
group by a.id,a.SName,a.ClsNo,a.Score
having count(b.id)<2
order by a.ClsNo,a.Score desc

方法二：

select *
from Table1 a
where 2>(select count(*) from Table1 where ClsNo=a.ClsNo and Score>a.Score)
order by a.ClsNo,a.Score desc

 

方法三：
select *
from Table1 a
where id in (select id from Table1 where ClsNo=a.ClsNo order by Score desc limit 2)
order by a.ClsNo,a.Score desc

 

方法....

这里列出了多种SQL语句的实现方法，有些是MySQL特有的(Limit, 其它数据库可根据实际更改，比如oracle的rownum,MS SQL SERVER 的 top,..)，有时是SQL标准支持的。但效率上和应用的场合或许不同。具体应用时可根据实际表中的记录情况，索引情况进行选择。

 

 

特例 N=1 ,即取最大的/最小的一条记录。
+----+------+------+-----+
| id |SName |ClsNo |Score|
+----+------+------+-----+
|  5 |EEEE  |  C1  | 95  |
|  7 |GGGG  |  C2  | 87  |
| 14 |NNNN  |  C3  | 99  |
+----+------+------+-----+


select * 
from Table1 a
where not exists (select 1 from Table1 where ClsNo=a.ClsNo and Score>a.Score);

select a.* 
from Table1 a inner join (select ClsNo, max(Score) as mScore from Table1 group by ClsNo) b
 on a.ClsNo=b.ClsNo and a.Score=b.Score

select *
from (select * from Table1 order by Score desc) t
group by ClsNo