代码之家 › 专栏 › 技术社区 › Stefan0309

SQL Pivot选择多行

unpivot pivot tsql sql-server sql

1

Stefan0309 · 技术社区 · 6 年前

这是我的桌子:

我想创建如下轴:

我写了一些查询,但我只得到对角线上的名字:

(
select 
[1] as Client0, 
[2] as Client1, 
[3] as Client2, 
[4] as Client3,
[5] as Client4
from 
(
select 
    rc.DateCreated, 
    gd.Name,
    DENSE_RANK() over (order by gd.ID_TableGD) as colnum
from TableGD gd
inner join TableRC rc ON gd.ID_TableGD = rc.ID_TableRC
WHERE gd.ID_TableGD IN (962,1029,1024) 
AND gd.Active = 1
) as t
pivot
(
MAX(Name)
for colnum in 
(
[1],
[2],
[3],
[4],
[5]

)
) as pvt)

这是我的第一个轴心,我不知道我是否需要把马克斯(姓名)?

2 回复 | 直到 6 年前

1

Fahmi 6 年前

选择时使用最大时间,如下所示:

select 
max([1]) as Client0, 
max([2]) as Client1, 
max([3]) as Client2, 
max([4]) as Client3,
max([5]) as Client4
from 
(
select 
    rc.DateCreated, 
    gd.Name,
    DENSE_RANK() over (order by gd.ID_TableGD) as colnum
from TableGD gd
inner join TableRC rc ON gd.ID_TableGD = rc.ID_TableRC
WHERE gd.ID_TableGD IN (962,1029,1024) 
AND gd.Active = 1
) as t
pivot
(
MAX(Name)
for colnum in 
(
[1],
[2],
[3],
[4],
[5]

)
) as pvt

2

D-Shih 6 年前

我会用 CROSS APPLY 具有 Value 去做那件事。

unpivot 让你的 DateCreated Name 在一列中,这是您的预期结果。

让结果像这样。

SELECT 
  v.*
FROM T t1 CROSS APPLY(
    VALUES 
    (Name,colnum,1),
    (CAST(DateCreated AS VARCHAR(50)),colnum,2)
) v (val,colnum,grp)

:

|        val | colnum | grp |
|------------|--------|-----|
|    ClientA |      1 |   1 |
| 2018-08-21 |      1 |   2 |
|    ClientB |      2 |   1 |
| 2018-08-22 |      2 |   2 |
|    ClientC |      3 |   1 |
| 2018-08-23 |      3 |   2 |

下一步您可以尝试使用 Pivot

CREATE TABLE T(
   Name VARCHAR(50),
   DateCreated DATE,
   colnum INT
);
INSERT INTO T VALUES('ClientA','2018-08-21',1);
INSERT INTO T VALUES('ClientB','2018-08-22',2);
INSERT INTO T VALUES('ClientC','2018-08-23',3);

查询1 :

SELECT 
    MAX(CASE WHEN v.colnum = 1 THEN v.val end) client0,
    MAX(CASE WHEN v.colnum = 2 THEN v.val end) client1,
    MAX(CASE WHEN v.colnum = 3 THEN v.val end) client2,
    MAX(CASE WHEN v.colnum = 4 THEN v.val end) client3,
    MAX(CASE WHEN v.colnum = 5 THEN v.val end) client4
FROM T t1 CROSS APPLY(
    VALUES 
    (Name,colnum,1),
    (CAST(DateCreated AS VARCHAR(50)),colnum,2)
) v (val,colnum,grp)
group by grp

Results :

|    client0 |    client1 |    client2 | client3 | client4 |
|------------|------------|------------|---------|---------|
|    ClientA |    ClientB |    ClientC |  (null) |  (null) |
| 2018-08-21 | 2018-08-22 | 2018-08-23 |  (null) |  (null) |

CROSS APPLY ... Value

你可以用 CONVERT 或 FORMAT

此示例用于 FORMAT http://sqlfiddle.com/#!18/12ae9e/3

尽管 format 使用起来比 Convert