SQL Server组的列行（支点？）

Question

我已经在SQL Server中得到这个结果数据2008 R2

id  Size Acted Sum  Avg1 Avg2 A1  A2  A3 
1  3921 39  690  17.69 0.18 NULL  NULL NULL 
40  11979 301  5944.26 19.75 0.5 10000.00 2000.00 1000.00 
41  11714 289  5060 17.51 0.43 10000.00 3000.00 2000.00 
42  11599 265  4107.98 15.5 0.35 10000.00 5000.00 500.00 

而且我想根据ID，所以我将收到的列移动到行此结果：

id1  id40  id41   id42 
1   40   41   42 
3921  11979  11714  11599 
39   301  289   265 
690  5944  5060   4107 
17.69  19.75  17.51  15.5 
0.18  0.5  0.43   0.35 
      10000.00 2000.00  1000.00 
      10000.00 3000.00  2000.00 
      10000.00 5000.00  500.00 

有没有办法做到这一点？ 我试过支点，但据我所知，我只能在这种情况下转换1列，并没有多少需要。

Answer 1

为了得到这个结果，你会先将unpivot的数据从列到行，然后应用PIVOT函数。

由于您使用的是SQL Server 2008，因此您可以使用CROSS APPLY和VALUES来反转数据。这将从您的众多列中获取值并将它们转换为行：

select 'id'+cast(t.id as varchar(10)) p_id, 
    c.col, 
    c.value, 
    c.sort_order 
from yourtable t 
cross apply 
(
    values 
    (1, 'id', id), 
    (2, 'size', size), 
    (3, 'acted', acted), 
    (4, 'sum', sum), 
    (5, 'avg1', avg1), 
    (6, 'avg2', avg2), 
    (7, 'a1', a1), 
    (8, 'a2', a2), 
    (9, 'a3', a3) 
) c (sort_order, col, value) 

请参阅SQL Fiddle with Demo。一旦数据未经转化，您就可以使用id值的新列进行调整。因此，完整的代码是：

select col, 
    id1, 
    id40, 
    id41, 
    id42 
from 
(
    select 'id'+cast(t.id as varchar(10)) p_id, 
    c.col, 
    c.value, 
    c.sort_order 
    from yourtable t 
    cross apply 
    (
    values 
     (1, 'id', id), 
     (2, 'size', size), 
     (3, 'acted', acted), 
     (4, 'sum', sum), 
     (5, 'avg1', avg1), 
     (6, 'avg2', avg2), 
     (7, 'a1', a1), 
     (8, 'a2', a2), 
     (9, 'a3', a3) 
) c (sort_order, col, value) 
) src 
pivot 
(
    max(value) 
    for p_id in (id1, id40, id41, id42) 
) piv 
order by sort_order; 

请参阅SQL Fiddle with Demo。

如果您不能使用CROSS APPLY和VALUES，那么这也可以做，使用UNPIVOT功能：

select col, 
    id1, id40, id41, id42 
from 
(
    select 'id'+cast(id_piv as varchar(10)) id, 
    col, 
    value, 
    case col 
     when 'id' then 1 
     when 'size' then 2 
     when 'acted' then 3 
     when 'sum' then 4 
     when 'avg1' then 5 
     when 'avg2' then 6 
     when 'a1' then 7 
     when 'a2' then 8 
     when 'a3' then 9 end sort_order 
    from 
    (
    select id id_piv, 
     cast(id as numeric(10, 2)) id, 
     cast(size as numeric(10, 2)) size, 
     cast(acted as numeric(10, 2)) acted, 
     sum, avg1, avg2, A1, A2, A3 
    from yourtable 
) d 
    unpivot 
    (
    value 
    for col in (id, size, acted, sum, avg1, avg2, a1, a2, a3) 
    ) unpiv 
) src 
pivot 
(
    max(value) 
    for id in (id1, id40, id41, id42) 
) piv 
order by sort_order; 

见SQL Fiddle with Demo

最后，如果你将有数目不详的id值要转换为列，那么您将需要使用动态sql：

DECLARE @cols AS NVARCHAR(MAX), 
    @query AS NVARCHAR(MAX) 

select @cols = STUFF((SELECT ',' + QUOTENAME('id'+cast(id as varchar(10))) 
        from yourtable 
        group by id 
        order by id 
      FOR XML PATH(''), TYPE 
      ).value('.', 'NVARCHAR(MAX)') 
     ,1,1,'') 

set @query = 'SELECT col, ' + @cols + ' 
       from 
      (
       select ''id''+cast(t.id as varchar(10)) p_id, 
        c.col, 
        c.value, 
        c.sort_order 
       from yourtable t 
       cross apply 
       (
        values 
        (1, ''id'', id), 
        (2, ''size'', size), 
        (3, ''acted'', acted), 
        (4, ''sum'', sum), 
        (5, ''avg1'', avg1), 
        (6, ''avg2'', avg2), 
        (7, ''a1'', a1), 
        (8, ''a2'', a2), 
        (9, ''a3'', a3) 
       ) c (sort_order, col, value) 
      ) x 
      pivot 
      (
       max(value) 
       for p_id in (' + @cols + ') 
      ) p 
      order by sort_order' 

execute(@query) 

见SQL Fiddle with Demo

所有版本的结果：

| COL | ID1 | ID40 | ID41 | ID42 | 
---------------------------------------------- 
| id |  1 |  40 | 41 |  42 | 
| size | 3921 | 11979 | 11714 | 11599 | 
| acted |  39 |  301 | 289 |  265 | 
| sum | 690 | 5944.26 | 5060 | 4107.98 | 
| avg1 | 17.69 | 19.75 | 17.51 | 15.5 | 
| avg2 | 0.18 |  0.5 | 0.43 | 0.35 | 
| a1 | (null) | 10000 | 10000 | 10000 | 
| a2 | (null) | 2000 | 3000 | 5000 | 
| a3 | (null) | 1000 | 2000 |  500 |