Question

我有表＆＃34; table1＆＃34;，其中包含以下行：

ColA    ColB
-------------
 p4      x1
 p4      z
 p4      z
 p5      s
 p5      t
 p5      y1
 p8      x1
 p8      y1
 p8      x1
 p9      x1
 p10     z1
 p1      z1
 p2      z1
 p3      z1

现在我需要显示一个这样的表:(如预期的那样）

ColA    ColB    Result
----------------------
 p4      x1       0
 p8      x1       0
 p9      x1       0
 p5      y1       0
 p8      y1       0
 p1      z1       0
 p10     z1       0
 p2      z1       0
 p3      z1       0

我正在撰写查询：

with cte 
as
(
    select colb,cola,dense_rank() over(partition by colb order by cola asc) tem
    from table1
)
select s.cola, s.colb, 0 as result into tem_table
from table1 s
inner join cte c on c.colb = s.colb
where tem > 1
group by s.cola,s.colb
order by s.colb;

但我不想在上面的查询中使用内连接，因为它具有较低的性能。所以我在尝试查询：

with cte as 
(
    select colb,cola,dense_rank() over(partition by colb order by cola asc) tem
    from table1
) 
select cola,colb,0 as result into tem_table from cte
where tem > 1 
group by cola,colb 
order by colb;

这给出了一个输出:(这是不正确的）

ColA    ColB    Result
----------------------
 p8      x1      0      
 p9      x1      0
 p8      y1      0
 p10     z1      0
 p2      z1      0
 p3      z1      0

如何在CTE中不使用内部联接的情况下获得预期的输出，因为它是自己加入的更多时间执行。

Answer 1

SELECT  colA, colB, 0
FROM    (
        SELECT  colA, colB, COUNT(*) OVER (PARTITION BY colB) cnt
        FROM    (
                SELECT  DISTINCT colA, colB
                FROM    mytable
                ) q
        ) q
WHERE   cnt > 1

PostgreSQL中公共表表达式内部联接

1 个答案: