一个查询排列问题

[timego]

现在有一个表，现在有两个字段一个ID，一个组名（里面有10来个组吧）。我要每组的前5个ID。这话话要怎么写啊？

[九斤半]

参考：

create table A(Score int ,Shift varchar(10))
insert A
select 20,'B' union all
select 30,'B' union all
select 10,'B' union all
select 78,'A' union all
select 19,'A' union all
select 5 ,'A' union all
select 40,'C' union all
select 25,'C' union all
select 6 ,'C'
--select * from A

create table B(Score int,Shift varchar(10))
insert B select 20,'B'
--select * from B

select * from A i
where score in
(
select top 2 score from A where Shift=i.Shift and score not in(select score from B where Shift=A.Shift)
)

drop table A,B

[子陌红尘]

select
a.*
from
表 a
where
a.ID in(select top 5 ID from 表 where 组名=a.组名 order by ID)

[dawugui]

按如下表取某个字段(ID)分类后同时按时间(rdate)升序、降序前三条记录。

if object_id('tempdb..#tab') is not null
drop table #tab
GO
create table #tab (
[id] [char](10),[age] [int],[rdate] [datetime])

insert into #tab(id,age,rdate) values('a' , 1 , '2006-01-01')
insert into #tab(id,age,rdate) values('a' , 2 , '2006-01-02')
insert into #tab(id,age,rdate) values('a' , 3 , '2006-01-03')
insert into #tab(id,age,rdate) values('a' , 4 , '2006-01-04')
insert into #tab(id,age,rdate) values('a' , 5 , '2006-01-05')
insert into #tab(id,age,rdate) values('a' , 6 , '2006-01-06')
insert into #tab(id,age,rdate) values('b' , 1 , '2006-02-01')
insert into #tab(id,age,rdate) values('b' , 2 , '2006-02-02')
insert into #tab(id,age,rdate) values('b' , 3 , '2006-02-03')
insert into #tab(id,age,rdate) values('b' , 4 , '2006-02-04')
insert into #tab(id,age,rdate) values('c' , 1 , '2006-03-01')
insert into #tab(id,age,rdate) values('c' , 2 , '2006-03-02')
insert into #tab(id,age,rdate) values('c' , 3 , '2006-03-03')
insert into #tab(id,age,rdate) values('d' , 1 , '2006-04-01')
insert into #tab(id,age,rdate) values('d' , 2 , '2006-04-02')
insert into #tab(id,age,rdate) values('e' , 1 , '2006-05-01')

--按时间rdate升序取前三条
select * from #tab t
where rdate in
(
select top 3 rdate from #tab where id=t.id order by rdate
)

id age rdate
---------- ----------- ------------------------------------------------------
a 1 2006-01-01 00:00:00.000
a 2 2006-01-02 00:00:00.000
a 3 2006-01-03 00:00:00.000
b 1 2006-02-01 00:00:00.000
b 2 2006-02-02 00:00:00.000
b 3 2006-02-03 00:00:00.000
c 1 2006-03-01 00:00:00.000
c 2 2006-03-02 00:00:00.000
c 3 2006-03-03 00:00:00.000
d 1 2006-04-01 00:00:00.000
d 2 2006-04-02 00:00:00.000
e 1 2006-05-01 00:00:00.000

(所影响的行数为 12 行)

--按时间rdate降序取前三条
select * from #tab t
where rdate in
(
select top 3 rdate from #tab where id=t.id order by rdate desc
)
order by id , rdate desc

id age rdate
---------- ----------- ------------------------------------------------------
a 6 2006-01-06 00:00:00.000
a 5 2006-01-05 00:00:00.000
a 4 2006-01-04 00:00:00.000
b 4 2006-02-04 00:00:00.000
b 3 2006-02-03 00:00:00.000
b 2 2006-02-02 00:00:00.000
c 3 2006-03-03 00:00:00.000
c 2 2006-03-02 00:00:00.000
c 1 2006-03-01 00:00:00.000
d 2 2006-04-02 00:00:00.000
d 1 2006-04-01 00:00:00.000
e 1 2006-05-01 00:00:00.000
(所影响的行数为 12 行)


--上面包含了总数不到3个的记录(即id为d,e的数据)，如果要取消它们，以升序为例(降序同理)
select * from #tab m
where rdate in
(
select top 3 rdate from
(
select * from #tab t
where id in
(
select id from #tab group by id having(count(*)) >= 3
)
) n
where m.id = n.id order by rdate
)

id age rdate
---------- ----------- ------------------------------------------------------
a 1 2006-01-01 00:00:00.000
a 2 2006-01-02 00:00:00.000
a 3 2006-01-03 00:00:00.000
b 1 2006-02-01 00:00:00.000
b 2 2006-02-02 00:00:00.000
b 3 2006-02-03 00:00:00.000
c 1 2006-03-01 00:00:00.000
c 2 2006-03-02 00:00:00.000
c 3 2006-03-03 00:00:00.000
(所影响的行数为 9 行)

--在上面的例中我们发现rdate都是不相同的，如果相同怎么办？
--例如id=a，第三条，第四条rdate相同，都为2006-01-03。
id age rdate
---------- ----------- ------------------------------------------------------
a 1 2006-01-01 00:00:00.000
a 2 2006-01-02 00:00:00.000
a 3 2006-01-03 00:00:00.000
a 4 2006-01-03 00:00:00.000
a 5 2006-01-05 00:00:00.000
a 6 2006-01-06 00:00:00.000
b 1 2006-02-01 00:00:00.000
b 2 2006-02-02 00:00:00.000
b 3 2006-02-03 00:00:00.000
b 4 2006-02-04 00:00:00.000
c 1 2006-03-01 00:00:00.000
c 2 2006-03-02 00:00:00.000
c 3 2006-03-03 00:00:00.000
d 1 2006-04-01 00:00:00.000
d 2 2006-04-02 00:00:00.000
e 1 2006-05-01 00:00:00.000

--如果想把第三、四都取出来，使用上面的语句即可。如果只取一条(只取第三不取第四)则要使用临时表了。
if object_id('tempdb..#tab') is not null
drop table #tab
GO
if object_id('tempdb..#temp') is not null
drop table #temp
GO
create table #tab (
[id] [char](10),[age] [int],[rdate] [datetime])

insert into #tab(id,age,rdate) values('a' , 1 , '2006-01-01')
insert into #tab(id,age,rdate) values('a' , 2 , '2006-01-02')
insert into #tab(id,age,rdate) values('a' , 3 , '2006-01-03')
insert into #tab(id,age,rdate) values('a' , 4 , '2006-01-03')
insert into #tab(id,age,rdate) values('a' , 5 , '2006-01-05')
insert into #tab(id,age,rdate) values('a' , 6 , '2006-01-06')
insert into #tab(id,age,rdate) values('b' , 1 , '2006-02-01')
insert into #tab(id,age,rdate) values('b' , 2 , '2006-02-02')
insert into #tab(id,age,rdate) values('b' , 3 , '2006-02-03')
insert into #tab(id,age,rdate) values('b' , 4 , '2006-02-04')
insert into #tab(id,age,rdate) values('c' , 1 , '2006-03-01')
insert into #tab(id,age,rdate) values('c' , 2 , '2006-03-02')
insert into #tab(id,age,rdate) values('c' , 3 , '2006-03-03')
insert into #tab(id,age,rdate) values('d' , 1 , '2006-04-01')
insert into #tab(id,age,rdate) values('d' , 2 , '2006-04-02')
insert into #tab(id,age,rdate) values('e' , 1 , '2006-05-01')

--按时间rdate升序取前三条(其他方法同上)
select id1=identity(int,1,1),* into #temp from #tab order by id , rdate /*(降序用rdate desc)*/
select * from #temp t
where id1 in
(
select top 3 id1 from #temp where id=t.id order by id1
)

[splory]

declare @sql varchar(8000)
set @sql=''
select @sql=@sql='select top 5 * from tb where 组名 ='+rtrim(a.组名)+' order by ID union all
' from (select distinct 组名 from tb) a
select @sql=left(@sql,len(@sql)-10)
exec(@sql)