SQL Server根据一列透视多个列

Question

我在sql server 2008R2中有以下源表和目标表.如何在TSQL中执行pivot(s)以从源到达目标.

SourceTbl

empId    empIndex    empState    empStDate    empEndDate
========================================================
10        1           AL          1/1/2012     12/1/2012
10        2           FL          2/1/2012     2/1/2013
15        1           FL          3/20/2012    1/1/2099

DestTbl

empId    empState1  empState1StDate    empState1EndDt    empState2  empState2StDate    empState2EndDt
=========================================================================================================
10        AL         1/1/2012           12/1/2012         FL         2/1/2012           2/1/2013
15        FL         3/20/2012          1/1/2099          NULL       NULL               NULL

希望empIndex以某种方式帮助支点.

Answer 1

由于您使用的是SQL Server,因此有几种不同的方法可以将行转换为列.您可以将聚合函数与CASE表达式一起使用:

select empid,
  max(case when empindex = 1 then empstate end) empState1,
  max(case when empindex = 1 then empStDate end) empStDate1,
  max(case when empindex = 1 then empEndDate end) empEndDate1,
  max(case when empindex = 2 then empstate end) empState2,
  max(case when empindex = 2 then empStDate end) empStDate2,
  max(case when empindex = 2 then empEndDate end) empEndDate2
from sourcetbl
group by empid;

请参阅SQL Fiddle with Demo.

如果你想使用旋转功能得到结果,那么我会建议先unpivoting列empState,empStDate并empEndDate因此你首先必须多行.您可以使用UNPIVOT函数或CROSS APPLY来转换代码所包含的数据:

select empid, col+cast(empindex as varchar(10)) col,  value
from sourcetbl
cross apply
(
  select 'empstate', empstate union all
  select 'empstdate', convert(varchar(10), empstdate, 120) union all
  select 'empenddate', convert(varchar(10), empenddate, 120)
) c (col, value);

见演示.一旦数据不透明,那么您可以应用PIVOT函数,因此最终代码将是:

select empid,
  empState1, empStDate1, empEndDate1,
  empState2, empStDate2, empEndDate2
from 
(
  select empid, col+cast(empindex as varchar(10)) col,  value
  from sourcetbl
  cross apply
  (
    select 'empstate', empstate union all
    select 'empstdate', convert(varchar(10), empstdate, 120) union all
    select 'empenddate', convert(varchar(10), empenddate, 120)
  ) c (col, value)
) d
pivot
(
  max(value)
  for col in (empState1, empStDate1, empEndDate1,
              empState2, empStDate2, empEndDate2)
) piv;

请参阅SQL Fiddle with Demo.

如果您的数量有限empindex,上述版本将会很好用,但如果没有,那么您可以使用动态SQL:

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

select @cols = STUFF((SELECT ',' + QUOTENAME(col+cast(empindex as varchar(10))) 
                    from SourceTbl
                    cross apply
                    (
                      select 'empstate', 1 union all
                      select 'empstdate', 2 union all
                      select 'empenddate', 3
                    ) c (col, so)
                    group by col, so, empindex
                    order by empindex, so
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

set @query = 'SELECT empid,' + @cols + ' 
            from 
            (
                select empid, col+cast(empindex as varchar(10)) col,  value
                from sourcetbl
                cross apply
                (
                  select ''empstate'', empstate union all
                  select ''empstdate'', convert(varchar(10), empstdate, 120) union all
                  select ''empenddate'', convert(varchar(10), empenddate, 120)
                ) c (col, value)
            ) x
            pivot 
            (
                max(value)
                for col in (' + @cols + ')
            ) p '

execute sp_executesql @query;

请参阅SQL Fiddle with Demo

您可以使用这些查询来INSERT INTO DestTbl,或者不是以这种格式存储数据,而是现在有一个查询来获得所需的结果.

这些查询以下列格式放置数据:

| EMPID | EMPSTATE1 | EMPSTDATE1 | EMPENDDATE1 | EMPSTATE2 | EMPSTDATE2 | EMPENDDATE2 |
---------------------------------------------------------------------------------------
|    10 |        AL | 2012-01-01 |  2012-12-01 |        FL | 2012-02-01 |  2013-02-01 |
|    15 |        FL | 2012-03-20 |  2099-01-01 |    (null) |     (null) |      (null) |