SQL Server - 加入动态透视表中的选择列答案

【问题标题】：SQL Server - Join a selection of columns from a Dynamically Pivoted TableSQL Server - 加入动态透视表中的选择列
【发布时间】：2018-05-08 09:57:05
【问题描述】：

我尝试动态旋转表 PMESATOVERVIEW 并将其连接到 PMEOBJECT 另一个表，代码如下。但是我想从透视表中删除某些列。

SQL 示例：http://rextester.com/DKHF52968

首先我将描述数据示例：

PMEOBJECT
ObjectId, Location,
1, North
2, South
3, West
4, East

并内连接该表的旋转版本：

PMESATOVERVIEW
ObjectId, AttributeCat, AttributeId, Value
1, WW, Surface, 40
1, XX, Date, 1-1-2017
1, WW, Energy Label, B
2, WW, Surface, 45
2, XX, Date, 3-5-2017
3, WW, Surface, 50
3, WW, Energy Label, A
4, WW, Surface, 50
4, XX, Date, 1-4-2018
4, WW, Energy Label, A+

使用以下代码：

DECLARE @colsValues 

AS NVARCHAR(max) = Stuff(
(SELECT DISTINCT ',' + 
    Quotename(fat.ATTRIBUTEID)
     FROM   PMESATOVERVIEW fat
     FOR xml path(''), type).value('.', 'NVARCHAR(MAX)'), 1, 1, '');

DECLARE @query AS NVARCHAR(max) =  
                                'SELECT    *
                                  FROM       (SELECT *
                                              FROM   PMESATOVERVIEW
                                              PIVOT(Max(VALUE)
                                              FOR ATTRIBUTEID IN ('+ 
@colsValues +')) piv) fat
                                  INNER JOIN PMEOBJECT tb
                                  ON          tb.OBJECTID = fat.OBJECTID';
EXECUTE(@query)

这段代码没有错误，但我得到的结果是：

PMESATOVERVIEW
ObjectId    AttributeCat    Date    Energy Label    Surface ObjectId    Location
1   WW  NULL        C       40      1   North
2   WW  NULL        B       40      2   West
3   WW  NULL        A+      45      3   South
4   WW  NULL        NULL    50      4   East
1   XX  1-1-2017    NULL    NULL    1   North
2   XX  3-4-2017    NULL    NULL    2   West
4   XX  1-5-2018    NULL    NULL    4   East

而我想要的是：

ObjectId    Date    Energy Label    Surface ObjectId    Location
1   1-1-2017    C       40  1   North
2   3-4-2017    B       40  2   West
3   NULL        A+      45  3   South
4   1-5-2018    NULL    50  4   East

SQL语句结果：http://rextester.com/DKHF52968

任何想法如何做到这一点？我尝试在 Pivot 语句中仅选择几列，但不知何故它无法识别 ObjectId、AttributeId 和 Value。

对此有何想法？

亲切的问候，伊戈尔

【问题讨论】：

标签： select dynamic pivot-table multiple-columns

【解决方案1】：

通过执行类似操作，您可以获得预期的输出。但是，可能有更好的方法来做到这一点。

DECLARE @cols AS NVARCHAR(max) = Stuff((SELECT DISTINCT ', ' + Quotename(AttributeId) 
         FROM   PMESATOVERVIEW       
         FOR xml path(''), type).value('.', 'NVARCHAR(MAX)'), 1, 1, ''); 

DECLARE @maxCols AS NVARCHAR(max) = Stuff((SELECT DISTINCT ', max(' + Quotename(AttributeId) + ') as ' + Quotename(AttributeId)
         FROM   PMESATOVERVIEW       
         FOR xml path(''), type).value('.', 'NVARCHAR(MAX)'), 1, 1, ''); 

DECLARE @query AS NVARCHAR(max) =  'SELECT     fat.ObjectId, '+@maxCols+',  max(tb.location) as Location 
                                    FROM       (SELECT ObjectId, '+ @cols +'
                                                FROM   PMESATOVERVIEW 
                                                PIVOT(Max(Value) FOR AttributeId IN ('+ @cols +')) piv) fat
                                    INNER JOIN PMEOBJECT tb 
                                    ON         tb.ObjectId = fat.ObjectId 
                                    group by   fat.ObjectId';

EXECUTE(@query)

输出

+----------+----------+--------------+---------+----------+
| ObjectId |   Date   | Energy Label | Surface | Location |
+----------+----------+--------------+---------+----------+
|        1 | 1-1-2017 | C            |      40 | North    |
|        2 | 3-4-2017 | B            |      40 | West     |
|        3 | NULL     | A+           |      45 | South    |
|        4 | 1-5-2018 | NULL         |      50 | East     |
+----------+----------+--------------+---------+----------+

在线演示：http://rextester.com/GVY57614

【讨论】：