Question

我有以下表格，其中包含SQL Server 2008 R2中的客户和购买数据：

客户

CustId   Last    First    Phone
   1     Doe     John    555-5555  
   2     Smith   Sally   444-4444
   3     Smith   Greg    222-1212

Order_Header

OrderId  CustId  Date  
 1001       3   07/08/2011
 1002       2   07/19/2011
 1003       2   03/12/2012
 1004       1   03/14/2012
 1005       3   03/20/2012
 1006       1   04/17/2012
 1007       2   06/04/2012
 1008       1   08/04/2011

Order_Lines

OrderId     Sequence    Item      Type    Manufacturer   Price
 1001          1        WIDGET     C         WidgCo      12.00
 1001          2        SWITCH     C        SwitchCo     10.00
 1002          1        RADIO      A        RadSupply    30.00
 1002          2        CRT        A        CRT&More     31.00
 1002          3        NCARD      G        iNetwork      5.00
 1003          1        SENSOR     E        Sensora       7.50
 1004          1        SENSOR     D        Sensora       6.00
 1005          1        WIDGET     C         WidgInc     11.50
 1006          1        RADIO      A        RadSupply    30.00
 1006          2        SCREEN     A        ScreensInc    2.00
 1007          1        ANTENNA    G       AntennasPlus   5.50
 1008          1        SWITCH     B       SwitchOutlet   6.00

我正在尝试创建一个查询，列出每个客户，他们的联系信息，以及他们在每个类型上花费的总额以及他们与每个制造商花费的总额。到目前为止，我已经尝试了各种不同的JOINS，但我似乎无法让它发挥作用我想找回这样的东西：

CustomerId  Last  First  Phone     TYPE  TOT_TYPE  
   1        Doe   John  555-5555    A      32.00
   1        Doe   John  555-5555    B      6.00  
   1        Doe   John  555-5555    D      6.00

我也不确定如何合并制造商总数。如何在同一查询中返回两个数据集？

Answer 1

听起来你想要PIVOT数据。这可以获得每个type的总计，然后每个总计manufacturer。 PIVOT有两种静态或动态方式。

静态（见SQL Fiddle with Demo）

SELECT custid, last, first, phone, type,
    IsNull([AntennaPlus], 0) as [AntennaPlus],
    IsNull([CRT&More], 0) as [CRT&More],
    IsNull([iNetwork], 0) as [iNetwork],
    IsNull([RadSupply], 0) as [RadSupply],
    IsNull([ScreensInc], 0) as [ScreensInc],
    IsNull([Sensora], 0) as [Sensora],
    IsNull([SwitchCo], 0) as [SwitchCo],
    IsNull([SwitchOutlet], 0) as [SwitchOutlet],
    IsNull([WidgCo], 0) as [WidgCo],  
    IsNull([AntennaPlus], 0)+ 
    IsNull([CRT&More], 0)+ 
    IsNull([iNetwork], 0)+ 
    IsNull([RadSupply], 0)+ 
    IsNull([ScreensInc], 0)+ 
    IsNull([Sensora], 0)+ 
    IsNull([SwitchCo], 0)+ 
    IsNull([SwitchOutlet], 0)+ 
    IsNull([WidgCo], 0) Total 
from 
 (
    SELECT c.custid, c.last, c.first, c.phone,
        ol1.type, 
        ol1.manufacturer,
        ol1.price
    from customers c
    inner join Order_Header oh
        on c.custid = oh.custid
    inner join Order_Lines ol1
        on oh.orderid = ol1.orderid
 ) x
 pivot 
 (
    sum(price)
    for manufacturer in ( [AntennaPlus], [CRT&More], [iNetwork], [RadSupply], [ScreensInc], [Sensora], [SwitchCo], [SwitchOutlet], [WidgCo])
 ) p

动态数据透视（参见SQL Fiddle with Demo）

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

select @cols = STUFF((SELECT distinct ', ' + QUOTENAME(manufacturer)
                    from Order_Lines
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

select @colsPivot = STUFF((SELECT distinct ',IsNull(' + QUOTENAME(manufacturer)+', 0) as ['+ manufacturer+ ']'
                    from Order_Lines
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

select @colsTotal = STUFF((SELECT distinct '+ IsNull(' + QUOTENAME(manufacturer)+', 0)'
                    from Order_Lines
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

set @query 
      = 'SELECT custid, last, first, phone, type,' + @colsPivot + ', '+ @colsTotal+' Total from 
         (
            SELECT c.custid, c.last, c.first, c.phone,
                ol1.type, 
                ol1.manufacturer,
                ol1.price
            from customers c
            inner join Order_Header oh
                on c.custid = oh.custid
            inner join Order_Lines ol1
                on oh.orderid = ol1.orderid
         ) x
         pivot 
         (
            sum(price)
            for manufacturer in (' + @cols + ')
         ) p '
execute(@query)

两者都应该提供相同的结果。区别在于您必须对所需的值进行硬编码的静态版本。动态版本在运行时获取列的值，因此如果制造商更改它将自动调整。

Answer 2

为什么这被贬低？这是一个有效的答案。问题似乎是希望不同行上的类型和制造商（它没有明确说明，但这是一个合理的解释）。没有评论的贬低肯定是不礼貌的。

您需要使用UNION执行此操作。一个提示是你的最终输出不适合你的问题。你没有制造商的地方。

select CustomerId, Last, First, Phone,
       Type, Manufacturer, TotPrice
from Customers C join
     ((select customerId, Type, NULL as Manufacturer, sum(Price) as TotPrice,
       from Order_Header oh join
            Order o
            on oh.OrderId = o.OrderId
       group by  customerId, Type
      ) union all
      ((select customerId, NULL as Type, Manufacturer, sum(Price) as TotPrice,
       from Order_Header oh join
            Order o
            on oh.OrderId = o.OrderId
       group by  customerId, Manufacturer
      )
     ) s
     on s.CustomerId = s.CustomerId

Answer 3

为什么需要它们在同一个数据集中？他们正在回答两个完全不同的问题，因此以任何方式组合它们都没有意义。如果可以将类型混合在一起，并且您希望TYPE列为TYPE / MANUFACTURER，那么它非常简单。如果您需要将类型放在列中，那么您可以将每个客户的所有内容放在一行中，然后按顺序排列。

如果您想要一个支点，请使用@ bluefeet的解决方案。否则，你可以考虑这个或戈登的版本。我认为这可能会表现得更好，但测试是有序的。

SELECT
   C.DesiredColumns,
   P.Category,
   P.Item
   P.TotalPrice
FROM
   dbo.Customers C
   INNER JOIN (
      SELECT
         H.CustomerId,
         V.Category,
         V.Item,
         Sum(L.Price) TotalPrice
      FROM
         dbo.Order_Header H
         INNER JOIN dbo.Order_Lines L
            ON H.OrderId = L.OrderId
         CROSS APPLY (
            VALUES
               ('Type', [Type]),
               ('Mfr', Manufacturer)
         ) V (Category, Item)
      GROUP BY
         H.CustomerId
         V.Category,
         V.Item
   ) P ON C.CustomerId = P.CustomerId

查询根据两个唯一维度返回聚合？

3 个答案: