T-SQL 动态数据透视

Posted 2023-03-29

【中文标题】T-SQL 动态数据透视

【英文标题】：T-SQL dynamic pivot

【发布时间】：2012-08-25 23:55:37

【问题描述】：

好的，我有一张像这样的表

ItemID | ColumnName | Value
1      | name       | Peter
1      | phone      | 12345678
1      | email      | [e-mail]
2      | name       | John
2      | phone      | 87654321
2      | email      | [e-mail]
3      | name       | Sarah
3      | phone      | 55667788
3      | email      | [e-mail]

现在我需要把它变成这样：

ItemID | name  | phone    | email
1      | Peter | 12345678 | [e-mail]
2      | John  | 87654321 | [e-mail]
3      | Sarah | 55667788 | [e-mail]

我一直在查看动态枢轴示例，但似乎无法将它们融入我的场景。

谁能帮忙？

【问题讨论】：

我在您的数据中看不到数据透视。您在第一个块中以一种奇怪的方式列出了常规列名和值，通常在第二个块中。

如果第一个块代表您的实际数据，那么为什么它会像这样存储而不是标准化？

为什么需要动态 PIVOT 来处理这些数据？这可以通过静态 PIVOT 来实现（检查我的答案）。是因为您希望 ColumnName 的值不是姓名、电话和电子邮件吗？

是的，列名和列数会改变。我应该提到这一点。

数据未标准化的相同原因。

【参考方案1】：

看看下面的例子

CREATE TABLE #Table (
        ID INT,
        ColumnName VARCHAR(250),
        Value VARCHAR(250)
)

INSERT INTO #Table SELECT 1,'name','Peter' 
INSERT INTO #Table SELECT 1,'phone','12345678' 
INSERT INTO #Table SELECT 1,'email','[e-mail]' 
INSERT INTO #Table SELECT 2,'name','John' 
INSERT INTO #Table SELECT 2,'phone','87654321' 
INSERT INTO #Table SELECT 2,'email','[e-mail]' 
INSERT INTO #Table SELECT 3,'name','Sarah' 
INSERT INTO #Table SELECT 3,'phone','55667788' 
INSERT INTO #Table SELECT 3,'email','[e-mail]' 

---I assumed your tablename as TESTTABLE--- 
DECLARE @cols NVARCHAR(2000) 
DECLARE @query NVARCHAR(4000) 

SELECT  @cols = STUFF(( SELECT DISTINCT TOP 100 PERCENT 
                                '],[' + t.ColumnName 
                        FROM    #Table AS t 
                        --ORDER BY '],[' + t.ID 
                        FOR XML PATH('') 
                      ), 1, 2, '') + ']' 

SELECT  @cols

SET @query = N'SELECT ID,'+ @cols +' FROM 
(SELECT t1.ID,t1.ColumnName , t1.Value FROM #Table AS t1) p 
PIVOT (MAX([Value]) FOR ColumnName IN ( '+ @cols +' )) 
AS pvt;' 

EXECUTE(@query)

DROP TABLE #Table

【讨论】：

非常感谢，这正是我想要的！

【参考方案2】：

试试这个：

SQL Server 2005+

 ;with 
        cte_name  as(select * from <table> where ColumnName='name'),
        cte_phone as(select * from <table> where ColumnName='phone'),
        cte_email as(select * from <table> where ColumnName='email')
  select n.ItemID,n.Value [Name],p.Value [Phone],e.Value [Email] 
  from  cte_name n
  join  cte_phone p
  on    n.ItemID=p.ItemID
  join  cte_email e
  on    n.ItemID=e.ItemID

SQL Fiddle Demo

【参考方案3】：

您不需要动态数据透视表，因为它将是一张不同的表格。只需执行以下操作：

name    phone   email
---------------------------------
Peter            
        123456
                 [e-mail]

查看SQL fiddle

SELECT DISTINCT u.ItemID, n.Value as 'name', p.Value as 'phone', e.Value as 'email'
FROM UserData u
INNER JOIN(
SELECT ItemID, Value 
FROM UserData WHERE ColumnName = 'name') n ON n.ItemID = u.ItemID
INNER JOIN(
SELECT ItemID, Value 
FROM UserData WHERE ColumnName = 'phone') p ON p.ItemID = u.ItemID
INNER JOIN(
SELECT ItemID, Value 
FROM UserData WHERE ColumnName = 'email') e ON e.ItemID = u.ItemID

【讨论】：

问题是 ColumnNames 是动态的。它们的名称和列数都会有所不同。

【参考方案4】：

这是我用于联系人列表的查询 :)

SELECT *
FROM
    (
    SELECT Contact_Id AS CT
          , [Age]
          , [Sex]
          , [State]
          , [Country]
          , [Keyword]
          , [Married]
          , [Kids]
          , [Car]
     FROM
         (SELECT c.PropertyName
               , c.ValueString
               , c.Contact_Id
          FROM
              ContactProfiles c) AS ctp
         PIVOT (max(ctp.ValueString) FOR PropertyName IN ([Age], [Sex], [State], [Country], [Keyword], [Married], [Kids], [Car])) AS PivotTable
         ) AS pvt

WHERE
    pvt.[Age] > 18
    AND (pvt.[State] = 'CA' OR pvt.[State] = 'NY')
    AND pvt.[Sex] = 'F'
    --*AND pvt.[Keyword] LIKE '%B;%'
    AND pvt.[Married] = 'True'
    AND pvt.[Kids] > 0

【参考方案5】：

你试过了吗：

SELECT ItemID, name, phone, email
FROM
(SELECT [ItemID] ,[ColumnName] ,[Value] FROM Item) Item
PIVOT (MAX(Value) FOR ColumnName IN (name, phone, email) ) as pvt 

【讨论】：

问题是 ColumnNames 是动态的。它们的名称和列数都会有所不同。