如何从同一个表中选择唯一的元组答案

【问题标题】：how to select unique tuples from same table如何从同一个表中选择唯一的元组
【发布时间】：2014-10-12 19:25:24
【问题描述】：

我想从

中找到所有同时注册两个或多个部分的不同学生对

enroll(sid*, grade, dname*, cno*, sectno*)

每个部分由 (dname*, cno*, sectno*) 唯一标识

【问题讨论】：

标签： sql database oracle

【解决方案1】：

您可以在标识一个部分的列上自行加入，并指定e1.sid < e2.sid 以仅获取每对学生一次。

select e1.sid, e2.sid from enroll e1
join enroll e2 on e1.dname = e2.dname
and e1.cno = e2.cno
and e1.secno = e2.sectno
where e1.sid < e2.sid
group by e1.sid, e2.sid
having count(*) > 1

【讨论】：

但我希望这对学生一起注册 2 个或更多部分。上面的查询给出了一对一起参加一个部分的人
@anvitapotluri 查看更新，而不是 distinct 使用 group by 和 having

【解决方案2】：

/*
with enroll as (
select 1 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 1 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 2 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 3 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 1 sid, 'dname' dname, 2 cno, 2 secno from dual union all
select 2 sid, 'dname' dname, 2 cno, 2 secno from dual
)
*/
select e1.sid, e2.sid from enroll e1
join enroll e2 on e1.dname = e2.dname
and e1.cno = e2.cno
and e1.secno = e2.secno
where e1.sid < e2.sid
group by e1.sid, e2.sid
having count(*) >= 2;

（但如果一个 sid 多次在同一部分注册，则此查询将无法正常工作）

【讨论】：

【解决方案3】：

with enroll as (
select 1 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 1 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 2 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 3 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 3 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 1 sid, 'dname' dname, 2 cno, 2 secno from dual union all
select 2 sid, 'dname' dname, 2 cno, 2 secno from dual
)
select rn,sid from (
select row_number() over (partition by sid order by dname) as rn ,sid
from enroll ) 
where rn>1

使用 row_number()

如果一个 sid 多次在同一部分注册，则此查询正常工作

如果需要，可以使用distinct

with enroll as (
select 1 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 1 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 2 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 3 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 3 sid, 'dname' dname, 1 cno, 1 secno from dual union all
select 1 sid, 'dname' dname, 2 cno, 2 secno from dual union all
select 4 sid, 'dname' dname, 2 cno, 2 secno from dual union all
select 2 sid, 'dname' dname, 2 cno, 2 secno from dual
)
select distinct  sid from (
select row_number() over (partition by sid order by dname) as rn ,sid
from enroll ) 
where rn>1

输出

>        SID
> ----------
>          1
>          2
>          3

【讨论】：