SQL Server组列作为行（枢轴？）

Question

我在SQL Server 2008 R2中得到了此结果数据

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

我想根据ID将列移动到行中，这样我将收到以下结果：

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？ 我尝试了数据透视，但据我所知，我只能转换1列，在这种情况下，转换的列数并不多。

Answer 1

为了获得此结果，您将首先要取消数据从列到行的PIVOT ，然后应用PIVOT函数。

由于使用的是SQL Server 2008，因此可以使用CROSS APPLY和VALUES取消数据透视。 这将从众多列中获取值并将其转换为行：

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)

请参阅带有演示的SQL Fiddle 。 一旦数据被取消透视，您就可以使用作为id值的新列进行透视。 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;

请参阅带有演示的SQL Fiddle 。

如果您不能使用CROSS APPLY和VALUES ，那么也可以使用UNPIVOT函数来完成：

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;

参见带有演示的SQL Fiddle

最后，如果您要转换为列的id值的数量未知，则需要使用动态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)

参见带有演示的SQL Fiddle

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 |