[转]SQL server 2005 PIVOT运算符的使用

本文转自：http://blog.csdn.net/robinson_0612/article/details/5385117

 

PIVOT，UNPIVOT运算符是SQL server 2005支持的新功能之一，主要用来实现行到列的转换。本文主要介绍PIVOT运算符的操作，以及如何实现动态PIVOT的行列转换。

       关于UNPIVOT及SQL server 2000下的行列转换请参照本人的其它文章。

一、PIVOT的语法
[sql] view plaincopyprint?
  1. SELECT  
  2.   [non-pivoted column], -- optional   
  3.   [additional non-pivoted columns], -- optional   
  4.   [first pivoted column],  
  5.   [additional pivoted columns]  
  6. FROM (  
  7.   SELECT query producing sql data for pivot  
  8.   -- select pivot columns as dimensions and   
  9.   -- value columns as measures from sql tables   
  10. ) AS TableAlias  
  11. PIVOT  
  12. (  
  13.  
  14.   <aggregation function>(columnfor aggregation or measure column) -- MIN,MAX,SUM,etc   
  15.  
  16.   FOR []  
  17.   IN (  
  18.     [first pivoted column], ..., [last pivoted column]  
  19.   )  
  20. ) AS PivotTableAlias  
  21. ORDERBY clause – optional  
SELECT 
  [non-pivoted column], -- optional 
  [additional non-pivoted columns], -- optional 
  [first pivoted column], 
  [additional pivoted columns] 
FROM ( 
  SELECT query producing sql data for pivot 
  -- select pivot columns as dimensions and 
  -- value columns as measures from sql tables 
) AS TableAlias 
PIVOT 
( 

  <aggregation function>(column for aggregation or measure column) -- MIN,MAX,SUM,etc 

  FOR [] 
  IN ( 
    [first pivoted column], ..., [last pivoted column] 
  ) 
) AS PivotTableAlias 
ORDER BY clause – optional

二、PIVOT的使用例子

1. 静态PIVOT的用法        为演示，从NorthWind数据库中提取一些记录生成新的Orders表，然后使用PIVOT将行转换到列。

[sql] view plaincopyprint?
  1. USE tempdb 
  2. GO 
  3. SELECTYEAR(OrderDate) AS [Year] 
  4.        ,CustomerID  
  5.        ,od.Quantity 
  6. INTO dbo.Orders        
  7. FROM NorthWind..Orders AS o 
  8.     JOIN NorthWind..[Order Details] AS od 
  9.         ON o.OrderID = od.OrderID 
  10. WHERE o.CustomerID IN ('BONAP','BOTTM','ANTON') 
  11. SELECT CustomerID 
  12.     ,[1996],[1997],[1998] 
  13. FROM dbo.Orders 
  14. PIVOT ( 
  15.        SUM(Quantity) 
  16.        FOR [Year] IN ([1996],[1997],[1998]) 
  17.    )x 
  18. /*              
  19. TSQL中pivot的结构： 
  20.     ●  用于生成pivot数据源的源表，作为一个输入表 
  21.     ●  pivot表 
  22.     ●  聚合列及透视列的选择 
  23.  
  24. TSQL中pivot的实现： 
  25. 1->上例中Orders表相当于是一个输入表。包含了CustomerID,[Year],Quantity 三个列。 
  26.    Year是透视列，用于生成维度。 
  27.    pivot首先将聚合列之外的列进行分组，并对其实现聚合。本列中则是对聚合列Quantity之外的列先实现分组， 
  28.    即对CustomerID，Year进行分组，并对其Quantity实现聚合，相当于先做如下处理： 
  29. */                  
  30. SELECT CustomerID 
  31.        ,[Year] 
  32.        ,SUM(Quantity) AS Total 
  33. FROM dbo.Orders 
  34. GROUPBY CustomerID 
  35.        ,[Year] 
  36. ORDERBY CustomerID             
  37.  
  38. /*  Result:         
  39. CustomerID Year        Total 
  40. ---------- ----------- -----------  
  41. ANTON      1996        24 
  42. ANTON      1997        295 
  43. ANTON      1998        40 
  44. BONAP      1996        181 
  45. BONAP      1997        486 
  46. BONAP      1998        313 
  47. BOTTM      1996        81 
  48. BOTTM      1997        454 
  49. BOTTM      1998        421 
  50. */ 
  51. /* 
  52. 2->pivot根据FOR [Year] IN子句中的值，在结果集中来建立对应的新列，本例中即是列，， 
  53.    对于新列，，中的取值，取中间结果集中与之相对应的值。 
  54.    如对于客户ANTON，1996列中的值就选择中间结果中对应的Total值，同理列中为。 
  55.    并将中间结果pivot表命名为x。 
  56.     
  57. 3->最外层的SELECT语句从pivot表生成最终结果,此处因Orders表仅有列，故直接将结果用一个SELECT返回，有嵌套的SELECT参照下例。 
  58.    
  59. --结果:    
  60. CustomerID 1996        1997        1998 
  61. ---------- ----------- ----------- -----------  
  62. ANTON      24          295         40 
  63. BONAP      181         486         313 
  64. BOTTM      81          454         421 
  65. */         以下是为输入表多于一列的例子,数据来源于SQL server 2005的AdventureWorks，其实现的原理同上。 
  66. SELECT *   
  67. FROM( 
  68.     SELECTYEAR(DueDate) [Year] 
  69.            ,CASEMONTH(DueDate) 
  70.             WHEN 1 THEN'January'  
  71.             WHEN 2 THEN'February' 
  72.             WHEN 3 THEN'March' 
  73.             WHEN 4 THEN'April' 
  74.             WHEN 5 THEN'May' 
  75.             WHEN 6 THEN'June' 
  76.             WHEN 7 THEN'July' 
  77.             WHEN 8 THEN'August' 
  78.             WHEN 9 THEN'September' 
  79.             WHEN 10 THEN'October' 
  80.             WHEN 11 THEN'November' 
  81.             WHEN 12 THEN'December' 
  82.            ENDas [Month] 
  83.            ,ProductID 
  84.            ,OrderQty 
  85.     FROM Production.WorkOrder 
  86. )WorkOrder 
  87.     PIVOT ( 
  88.            SUM(OrderQty) 
  89.            FOR [Month] IN ([January],[February],[March],[April],[May],[June],[July],[August],[September],[October],[November],[December]) 
  90.           )x 
  91. ORDERBY [Year], ProductID   
  92. --Result: 末尾部分省略  
  93. /*   
  94. Year        ProductID   January     February    March       April       May         June        July        August       
  95. ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- -----------   
  96. 2002        3           8480        16870       12960       9530        19390       14170       26200       35870              
  97. 2002        316         1842        3704        2910        2252        4738        3496        7624        10778           
  98. 2002        324         1842        3704        2910        2252        4738        3496        7546        10600             
  99. 2002        327         921         1852        1455        1126        2369        1748        3773        5300               
  100. 2002        328         414         1048        872         458         1272        992         1786        2632          
  101. */  
USE tempdb
GO
SELECT YEAR(OrderDate) AS [Year]
       ,CustomerID 
       ,od.Quantity
INTO dbo.Orders       
FROM NorthWind..Orders AS o
    JOIN NorthWind..[Order Details] AS od
        ON o.OrderID = od.OrderID
WHERE o.CustomerID IN ('BONAP','BOTTM','ANTON')
SELECT CustomerID
	,[1996],[1997],[1998]
FROM dbo.Orders
PIVOT (
	   SUM(Quantity)
	   FOR [Year] IN ([1996],[1997],[1998])
   )x
/*			   
TSQL中pivot的结构：
    ●  用于生成pivot数据源的源表，作为一个输入表
    ●  pivot表
    ●  聚合列及透视列的选择

TSQL中pivot的实现：
1->上例中Orders表相当于是一个输入表。包含了CustomerID,[Year],Quantity 三个列。
   Year是透视列，用于生成维度。
   pivot首先将聚合列之外的列进行分组，并对其实现聚合。本列中则是对聚合列Quantity之外的列先实现分组，
   即对CustomerID，Year进行分组，并对其Quantity实现聚合，相当于先做如下处理：
*/  			   
SELECT CustomerID
       ,[Year]
       ,SUM(Quantity) AS Total
FROM dbo.Orders
GROUP BY CustomerID
       ,[Year]
ORDER BY CustomerID			   

/*	Result:		   
CustomerID Year        Total
---------- ----------- -----------
ANTON      1996        24
ANTON      1997        295
ANTON      1998        40
BONAP      1996        181
BONAP      1997        486
BONAP      1998        313
BOTTM      1996        81
BOTTM      1997        454
BOTTM      1998        421
*/
/*
2->pivot根据FOR [Year] IN子句中的值，在结果集中来建立对应的新列，本例中即是列，，
   对于新列，，中的取值，取中间结果集中与之相对应的值。
   如对于客户ANTON，1996列中的值就选择中间结果中对应的Total值，同理列中为。
   并将中间结果pivot表命名为x。
   
3->最外层的SELECT语句从pivot表生成最终结果,此处因Orders表仅有列，故直接将结果用一个SELECT返回，有嵌套的SELECT参照下例。
  
--结果:  
CustomerID 1996        1997        1998
---------- ----------- ----------- -----------
ANTON      24          295         40
BONAP      181         486         313
BOTTM      81          454         421
*/         以下是为输入表多于一列的例子,数据来源于SQL server 2005的AdventureWorks，其实现的原理同上。
SELECT *  
FROM(
	SELECT YEAR(DueDate) [Year]
		   ,CASE MONTH(DueDate)
			WHEN 1 THEN 'January' 
			WHEN 2 THEN 'February'
			WHEN 3 THEN 'March'
			WHEN 4 THEN 'April'
			WHEN 5 THEN 'May'
			WHEN 6 THEN 'June'
			WHEN 7 THEN 'July'
			WHEN 8 THEN 'August'
			WHEN 9 THEN 'September'
			WHEN 10 THEN 'October'
			WHEN 11 THEN 'November'
			WHEN 12 THEN 'December'
		   END as [Month]
		   ,ProductID
		   ,OrderQty
	FROM Production.WorkOrder
)WorkOrder
    PIVOT (
           SUM(OrderQty)
           FOR [Month] IN ([January],[February],[March],[April],[May],[June],[July],[August],[September],[October],[November],[December])
          )x
ORDER BY [Year], ProductID  
--Result: 末尾部分省略
/*  
Year        ProductID   January     February    March       April       May         June        July        August      
----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- 
2002        3           8480        16870       12960       9530        19390       14170       26200       35870             
2002        316         1842        3704        2910        2252        4738        3496        7624        10778          
2002        324         1842        3704        2910        2252        4738        3496        7546        10600            
2002        327         921         1852        1455        1126        2369        1748        3773        5300              
2002        328         414         1048        872         458         1272        992         1786        2632         
*/

2. 动态PIVOT的使用

[sql] view plaincopyprint?
  1. USE AdventureWorks; 
  2. GO  
  3.  
  4. --第一种生成透视列的方法，使用了COALESCE来联接字符串  
  5. DECLARE @PivotColHeader VARCHAR(MAX)     
  6. SELECT @PivotColHeader = 
  7.     COALESCE(@PivotColHeader + ',[' + cast(Nameasvarchar) + ']', 
  8.     '[' + cast(Nameasvarchar) + ']')   --示例中Name转换为varchar或char类型，注意：在CAST 和CONVERT 中使用varchar 时，显示n的默认值为30  
  9. FROM Sales.SalesTerritory 
  10. GROUPBYName 
  11.  
  12. /* 
  13. --第二种生成透视列的方法，使用了FOR XML PATH方法  
  14. SELECT @PivotColHeader =  
  15.     STUFF( 
  16.     (      
  17.           SELECTDISTINCT',[' + cast(Nameasvarchar) + ']' 
  18.           FROM Sales.SalesTerritory 
  19.           FOR XML PATH('') 
  20.     ), 
  21.     1,1,'') 
  22. */ 
  23.  
  24. DECLARE @PivotTableSQL NVARCHAR(MAX) 
  25. SET @PivotTableSQL = N' 
  26.     SELECT * 
  27.     FROM ( 
  28.         SELECTYEAR(H.OrderDate) [Year] 
  29.             ,T.Name 
  30.             ,H.TotalDue 
  31.         FROM Sales.SalesOrderHeader H 
  32.             LEFTJOIN Sales.SalesTerritory T 
  33.                 ON H.TerritoryID = T.TerritoryID 
  34.     )AS PivotData 
  35.         PIVOT( 
  36.             SUM(TotalDue) 
  37.             FORNameIN ( 
  38.                          ' + @PivotColHeader + ' 
  39.                          ) 
  40.                ) AS x '                             
  41. EXECUTE sp_executesql @PivotTableSQL     
  42.  
  43. --Result:部分结果省略  
  44. /* 
  45. Year        Australia             Canada                Central               France                Germany               Northeast              
  46. ----------- --------------------- --------------------- --------------------- --------------------- --------------------- ---------------------   
  47. 2001        1446497.1744          2173647.1453          1263884.1024          199531.723            262752.4184           754833.2045            
  48. 2002        2380484.8387          7215430.5017          3518185.4756          1717145.7439          575960.0974           3275322.1694           
  49. 2003        4547123.2777          8186021.9178          4015356.874           4366078.3475          2714826.4297          3833030.25            
  50. 2004        3823410.2386          3926712.8926          1771532.7396          2853948.6596          2386224.5508          1406555.6861          
  51.  
  52. */     
USE AdventureWorks;
GO 

--第一种生成透视列的方法，使用了COALESCE来联接字符串
DECLARE @PivotColHeader VARCHAR(MAX)    
SELECT @PivotColHeader =
    COALESCE(@PivotColHeader + ',[' + cast(Name as varchar) + ']',
    '[' + cast(Name as varchar) + ']')   --示例中Name转换为varchar或char类型，注意：在CAST 和CONVERT 中使用varchar 时，显示n的默认值为30
FROM Sales.SalesTerritory
GROUP BY Name

/*
--第二种生成透视列的方法，使用了FOR XML PATH方法
SELECT @PivotColHeader = 
    STUFF(
    (     
          SELECT DISTINCT ',[' + cast(Name as varchar) + ']'
          FROM Sales.SalesTerritory
          FOR XML PATH('')
    ),
    1,1,'')
*/

DECLARE @PivotTableSQL NVARCHAR(MAX)
SET @PivotTableSQL = N'
    SELECT *
    FROM (
        SELECT YEAR(H.OrderDate) [Year]
			,T.Name
			,H.TotalDue
        FROM Sales.SalesOrderHeader H
			LEFT JOIN Sales.SalesTerritory T
			    ON H.TerritoryID = T.TerritoryID
    )AS PivotData
        PIVOT(
            SUM(TotalDue)
            FOR Name IN (
                         ' + @PivotColHeader + '
                         )
               ) AS x '                            
EXECUTE sp_executesql @PivotTableSQL    

--Result:部分结果省略
/*
Year        Australia             Canada                Central               France                Germany               Northeast             
----------- --------------------- --------------------- --------------------- --------------------- --------------------- --------------------- 
2001        1446497.1744          2173647.1453          1263884.1024          199531.723            262752.4184           754833.2045           
2002        2380484.8387          7215430.5017          3518185.4756          1717145.7439          575960.0974           3275322.1694          
2003        4547123.2777          8186021.9178          4015356.874           4366078.3475          2714826.4297          3833030.25           
2004        3823410.2386          3926712.8926          1771532.7396          2853948.6596          2386224.5508          1406555.6861         

*/

   对该动态pivot增加汇总列

[sql] view plaincopyprint?
  1. DECLARE @PivotColHeader VARCHAR(MAX) 
  2. DECLARE @TotalCol VARCHAR(MAX) 
  3.  
  4. SELECT @PivotColHeader =                              --使用COALESCE函数生成列标题  
  5.     COALESCE(@PivotColHeader + ',[' + cast(Nameasvarchar) + ']', 
  6.     '[' + cast(Nameasvarchar) + ']') 
  7.     , 
  8.     @TotalCol = COALESCE(@TotalCol + ', SUM([' + cast(Nameasvarchar) + ']) AS [' + cast(Nameasvarchar) + ']' 
  9.      ,'SUM([' + cast(Nameasvarchar) + ']) AS [' + cast(Nameasvarchar) + ']')     --使用COALESCE函数生成汇总字符串  
  10. FROM Sales.SalesTerritory 
  11.  
  12. DECLARE @PivotTableSQL NVARCHAR(MAX) 
  13. SET @PivotTableSQL = N' 
  14.     SELECT * 
  15.     FROM ( 
  16.         SELECTCAST(YEAR(H.OrderDate) ASCHAR(4)) [Year] 
  17.         ,T.Name 
  18.         ,H.TotalDue 
  19.         FROM Sales.SalesOrderHeader H 
  20.         LEFTJOIN Sales.SalesTerritory T 
  21.            ON H.TerritoryID = T.TerritoryID 
  22.     )AS PivotData 
  23.         PIVOT( 
  24.             SUM(TotalDue) 
  25.             FORNameIN ( 
  26.              ' + @PivotColHeader + ' 
  27.                    ) 
  28.             ) AS x    
  29.     UNION  
  30.     SELECT''GrandTotal'', ' + @TotalCol + ' 
  31.     FROM ( 
  32.     SELECTCAST(YEAR(H.OrderDate) ASCHAR(4)) [Year] 
  33.         ,T.Name 
  34.         ,H.TotalDue 
  35.     FROM Sales.SalesOrderHeader H 
  36.         LEFTJOIN Sales.SalesTerritory T 
  37.         ON H.TerritoryID = T.TerritoryID 
  38.     ) AS PivotData                                                   
  39.         PIVOT( 
  40.             SUM(TotalDue) 
  41.             FORNameIN ( 
  42.                 ' + @PivotColHeader + ' 
  43.                    ) 
  44.              ) AS y '    
  45. --PRINT  @PivotTableSQL                                          
  46. EXECUTE sp_executesql @PivotTableSQL  
  47.  
  48. --Result:部分结果省略  
  49. /* 
  50. Year       Australia             Canada                Central               France                Germany               Northeast             Northwest              
  51. ---------- --------------------- --------------------- --------------------- --------------------- --------------------- --------------------- ---------------------   
  52. 2001       1446497.1744          2173647.1453          1263884.1024          199531.723            262752.4184           754833.2045           2703481.7947           
  53. 2002       2380484.8387          7215430.5017          3518185.4756          1717145.7439          575960.0974           3275322.1694          5651688.6685           
  54. 2003       4547123.2777          8186021.9178          4015356.874           4366078.3475          2714826.4297          3833030.25            7494658.0357           
  55. 2004       3823410.2386          3926712.8926          1771532.7396          2853948.6596          2386224.5508          1406555.6861          4952772.2793           
  56. GrandTotal 12197515.5294         21501812.4574         10568959.1916         9136704.474           5939763.4963          9269741.31            20802600.7782          
  57. */      
DECLARE @PivotColHeader VARCHAR(MAX)
DECLARE @TotalCol VARCHAR(MAX)

SELECT @PivotColHeader =                              --使用COALESCE函数生成列标题
    COALESCE(@PivotColHeader + ',[' + cast(Name as varchar) + ']',
    '[' + cast(Name as varchar) + ']')
    ,
    @TotalCol = COALESCE(@TotalCol + ', SUM([' + cast(Name as varchar) + ']) AS [' + cast(Name as varchar) + ']'
     ,'SUM([' + cast(Name as varchar) + ']) AS [' + cast(Name as varchar) + ']')     --使用COALESCE函数生成汇总字符串
FROM Sales.SalesTerritory

DECLARE @PivotTableSQL NVARCHAR(MAX)
SET @PivotTableSQL = N'
    SELECT *
    FROM (
        SELECT CAST(YEAR(H.OrderDate) AS CHAR(4)) [Year]
		,T.Name
		,H.TotalDue
        FROM Sales.SalesOrderHeader H
		LEFT JOIN Sales.SalesTerritory T
		   ON H.TerritoryID = T.TerritoryID
    )AS PivotData
		PIVOT(
			SUM(TotalDue)
			FOR Name IN (
			 ' + @PivotColHeader + '
				   )
		    ) AS x   
    UNION 
    SELECT ''GrandTotal'', ' + @TotalCol + '
    FROM (
	SELECT CAST(YEAR(H.OrderDate) AS CHAR(4)) [Year]
		,T.Name
		,H.TotalDue
	FROM Sales.SalesOrderHeader H
	    LEFT JOIN Sales.SalesTerritory T
		ON H.TerritoryID = T.TerritoryID
	) AS PivotData													
		PIVOT(
			SUM(TotalDue)
			FOR Name IN (
				' + @PivotColHeader + '
				   )
		     ) AS y '   
--PRINT  @PivotTableSQL                                        
EXECUTE sp_executesql @PivotTableSQL 

--Result:部分结果省略
/*
Year       Australia             Canada                Central               France                Germany               Northeast             Northwest             
---------- --------------------- --------------------- --------------------- --------------------- --------------------- --------------------- --------------------- 
2001       1446497.1744          2173647.1453          1263884.1024          199531.723            262752.4184           754833.2045           2703481.7947          
2002       2380484.8387          7215430.5017          3518185.4756          1717145.7439          575960.0974           3275322.1694          5651688.6685          
2003       4547123.2777          8186021.9178          4015356.874           4366078.3475          2714826.4297          3833030.25            7494658.0357          
2004       3823410.2386          3926712.8926          1771532.7396          2853948.6596          2386224.5508          1406555.6861          4952772.2793          
GrandTotal 12197515.5294         21501812.4574         10568959.1916         9136704.474           5939763.4963          9269741.31            20802600.7782         
*/

生成汇总列的注意事项;     1->使用COALESCE函数生成列标题 。     2->使用COALESCE函数生成带有SUM求和函数并且指定了别名的字符串。     3->使用UNION对两个SELECT来实现联接。且将[Year]转换为字符串，因为YEAR(H.OrderDate)得值为 INT ，而''GrandTotal''为字符串，UNION 或UNION ALL使用时必须列的数量和类型相对应。

 

posted on 2013-06-04 15:27  freeliver54  阅读(273)  评论(0)  编辑  收藏  举报

导航