SQL服务器：转换行到列的无支点和分组

Question

本来我有一个SQL Server表如下图所示：

Result Type Type Order Type Value 
-------------------------------------- 
KM1 Color 1    Blue 
KM1 Shape 2    Square 
KM1 Size  3    10 
KM3 Color 1    Blue 
KM3 Shape 2    Square 
KM3 Size  3    10 
KM2 Color 1    Red 
KM2 Shape 2    Round 
KM2 Size  3    20 
KM2 Color 1    Blue 
KM2 Shape 2    Square 
KM2 Size  3    10 
KM2 Color 1    Blue 
KM2 Shape 2    Round 
KM2 Size  3    10 

和我预期的结果应该是像

Color Shape Size Result 
------------------------------------- 
Blue Square 10  KM1, KM3, KM2 
Red  Round 20  KM2 
Blue Round 10  KM2 

能这样做吗？我不认为这个关键点会有帮助，因为我对于多个数值组合有相同的结果。

Answer 1

这是两个级别的聚合。首先是描述每个结果。第二个是把钥匙放在一起。所以：

with t as (
     select result, 
      max(case when type = 'Color' then value end) as color, 
      max(case when type = 'Size' then value end) as size, 
      max(case when type = 'Shape' then value end) as shape 
     from t 
     group by result 
    ) 
select color, size, shape, 
     stuff((select ',' + t2.result 
       from t t2 
       where t2.color = t.color and t2.size = t.size and t2.shape = t.shape 
       for xml path ('') 
      ), 1, 1, '') as keys 
from (select distinct color, size, shape 
     from t 
    ) t;