SQL Server unpivot two columns

问题:

I'm trying to pivot a table to get 3 columns

my example table is like :

CREATE TABLE tbl1 (A1 int, cA1 int,A2 int, cA2 int,A3 int, cA3 int)
GO
INSERT INTO tbl1  VALUES (60,2,30,3,10,5);
GO

I am using the query below to get tthe results from two columns:

select A, value from tbl1
unpivot
(
value
for A in ([A1], [A2],[A3])
) un1;

The results are like :

A | value
--+-------
A1|60
A2|30
A3|10

but I want to add and second column with and the results to be like :

A | value1 | value2
--+--------+--------
A1| 60     | 2
A2| 30     | 3
A3| 10     | 5

Any Help??

回答1:

I would use APPLY:

select v.*
from tbl1 t cross apply
     (values ('A1', t.A1, t.cA1),
             ('A2', t.A2, t.cA2),
             ('A3', t.A3, t.cA3)
     ) v(A, value1, value2);

CROSS APPLY implements a lateral join. This is much more powerful than merely unpivoting data, although unpivoting data is one simple way to start learning about lateral joins.

回答2:

Another way with XML:

DECLARE @x xml = (SELECT * FROM tbl1 as t FOR XML AUTO, TYPE)

;WITH cte AS (
SELECT  CAST(t.c.query('local-name(.)') as nvarchar(10)) as [name],
        t.c.value('.','int') as [value],
        p.number as [pos]
FROM [master]..spt_values p
CROSS APPLY @x.nodes('/t[position()=sql:column("number")]/@*') as t(c)
WHERE p.[type] = 'p'
)

SELECT  c.[name] as A,
        c.[value] as value1,
        c1.[value] as value2
FROM cte c
INNER JOIN cte c1
    ON c1.[name] = N'c'+c.[name] and c.pos = c1.pos

Output:

A   value1  value2
A1  60      2
A2  30      3
A3  10      5