Sql 对列求和，对行进行计数，使用多个表一起显示数据

我有一个复杂的SQL问题，如下表和数据所示：

订单表：

OrderId    ClientId   CompanyId
--------------------------------
25          2           NULL
26          NULL        1 
27          3           NULL
28          3           NULL
29          NULL        3

Billing table                  Forwards table

OrderId    Amount PaidFull     OrderId    Amount  PaidFull
--------------------------     ---------------------------
25         $100    False         26        $25    False
26         $50     False         27        $50    False
27         $100    False         27        $50    False
28         $100    False         29        $50    False
29         $100    True


Client table                 Company table

ClientId     ClientName      CompanyId      CompanyName
-----------------------      ---------------------------
1            AClient          1              ACompany
2            BClient          2              BCompany
3            CClient          3              CCompany

我想在下面这样的表格中显示数据，每个公司或客户的总账单合计和行计数。每个公司或客户转发汇总和计数的行，并按照客户和公司名称的组合按字母顺序排列：

Billing  Qty   Forwards Qty   Client/Company Name
--------------------------------------------------
$50       1    $25       1     ACompany
$100      1    $0        0     BClient
$200      2    $100      2     CClient
$0        0    $50       1     CCompany

问题是，使用下面的SQL查询，我得到了重复的值，对账单表求和，因为它对账单表orderid29求和$100，尽管它标记为PaidFull=True。另外，它不是按客户表和公司表的组合按字母顺序排序，而是先将它们分开，然后再排序

Select 
    Sum(b.Amount) As TotalAmount, 
    Count(b.Amount) As Count1, 
    Sum(f.Amount) As TotalAmountf, 
    Count(f.Amount) As Count2,
    d.ClientName, c.CompanyName
From
    Orders a 
Left Join 
    Billing b On a.OrderId = b.OderId And b.PaidFull = 'False'
Left Join 
    Forwards f On a.OrderId = f.OderId And f.PaidFull = 'False'
Left Join 
    Client d On a.ClientId = d.ClientId
Left Join 
    Company c On a.CompanyId = c.CompanyId
Where 
    (b.FaidFull = 'False' Or f.PaidFull = 'False')
Group by 
    ClientName, CompanyName
Order by 
    ClientName, CompanyName

---

以下是您如何做到这一点：

select  (case when cmp.CompanyName is null then c.ClientName else cmp.CompanyName end) "Client/Company Name"
        , sum(b.Amount) Billing
        , max(b.billingCount) QTY
        , sum(f.Amount) Forwards
        , max(f.forwardsCount) QTY
from orders o 
join Clients c
    on c.ClientId = o.ClientId
join Company cmp
    on cmp.CompanyId = o.CompanyId
join ( select OrderId , sum(amount) amount , count(*) billingCount
       from Billing b
       group by orderId
     ) b
    on b.OrderId = o.OrderId
join ( select OrderId , sum(amount(*) amount , count(*) forwardsCount
       from forwards f
       group by OrderId
      ) f
    on f.OrderId = o.OrderId
group by (case when cmp.CompanyName is null then c.ClientName else cmp.CompanyName end)

---

虽然我不确定您的

账单

和

转发

的表结构，但实现这一点的最简单方法是使用CTE首先对这两个表中的所有值求和，然后在最终查询中使用这些值，如下所示：

WITH TotalBilling
AS
(
    SELECT
         OrderId
        ,SUM(Amount) Amount
    FROM
        Billing
    WHERE
        PaidFull = 'FALSE'
    GROUP BY
        OrderId
),

TotalForwards
AS
(
    SELECT
         OrderId
        ,SUM(Amount) Amount
    FROM
        Forwards
    WHERE
        PaidFull = 'FALSE'
    GROUP BY
        OrderId
)

SELECT
     SUM(B.Amount)                           AS Billing
    ,COUNT(B.OrderId)                        AS Qty
    ,SUM(F.Amount)                           AS Forwards
    ,COUNT(F.OrderId)                        AS Qty
    ,COALESCE(Cl.ClientName, Co.CompanyName) AS "Client / Company Name"
FROM
    Orders O
    LEFT JOIN Clients Cl
    ON O.ClientId = Cl.ClientId
    LEFT JOIN Companies Co
    ON O.CompanyId = Co.CompanyId
    LEFT JOIN TotalBilling B
    ON B.OrderId = O.OrderId
    LEFT JOIN TotalForwards F
    ON F.OrderId = O.OrderId
GROUP BY
    COALESCE(Cl.ClientName, Co.CompanyName)
ORDER BY
     COALESCE(Cl.ClientName, Co.CompanyName)

---

是一个SqlFiddle，显示了它的外观。

我不确定我是否理解

账单

和

转发

表-如何在转发表中为相同的

订单ID设置两个条目

（同样的情况会发生在账单上吗？可以想象，会出现什么情况，你的查询应该如何处理？你的答案是有效的。我只是在分组后加了一句，“Sum（B.Amount）>0或Sum（F.Amount）>0”为了不选择PaidFull=True的行。因为选择了PaidFull=True的行。为了回答您的问题，是的，账单和转发表中可能有多个相同的OrderID。从现在开始，这就可以工作了，我希望保持这种状态！我真的不确定您为什么需要它，因为它应该在CTEs.我在示例数据集中使用了额外的

HAVING

子句，它返回了相同的表。但是，假设您的生产数据需要它，我很高兴它能为您工作，只是不确定您为什么需要添加它。