在带有重置条件的 SQL 中运行总计

Question

我有一个数据，它结合了不同零件的库存、销售预测和未来供应数据。我把它结合起来给我一个表格，每个月给出每个部分的净数量。见下文：

Date	Part	Net Quantity
30/06/2021	A	1000
31/07/2021	A	-150
31/08/2021	A	-200
30/09/2021	A	-500
31/10/2021	A	-200
30/11/2021	A	-200
31/12/2021	A	50
30/06/2021	B	100
31/07/2021	B	-80
31/08/2021	B	20
30/09/2021	B	-30
31/10/2021	B	-35
30/11/2021	B	-40
31/12/2021	B	-150

我需要创建一个按零件编号分区的运行总计视图，如果低于 0，则在下个月重置。如果上个月的期末库存为负，则运行总计的起点需要为 0。

我的预期结果是这样的：

Date	Part	Net Quantity	Closing Inventory (Expected Outcome)
30/06/2021	A	1000	1000
31/07/2021	A	-150	850
31/08/2021	A	-200	650
30/09/2021	A	-500	150
31/10/2021	A	-200	-50
30/11/2021	A	-200	-200
31/12/2021	A	50	50
30/06/2021	B	100	100
31/07/2021	B	-80	20
31/08/2021	B	20	40
30/09/2021	B	-30	10
31/10/2021	B	-35	-25
30/11/2021	B	-40	-40
31/12/2021	B	-150	-150

我当前的代码是：

SELECT 
Date, 
Part, 
Net_Quantity, 
sum(Net_Quantity) over (partition by Part order by date) 'Closing_Inventory' 
FROM grouped

我不确定如何根据运行总计中的最后一个值创建条件，lag 函数无法查看运行总计的上一行。

Answer 1

这样的事情怎么办……

[不正确的答案已删除...]

编辑：

--Add a column...
ALTER TABLE PartsInventory ADD RowNumWithinPart int;

--Assign row numbers partitioned by Part...
UPDATE PartsInventory
SET RowNumWithinPart = x.rownum
FROM (
    SELECT ROW_NUMBER() OVER (PARTITION BY Part ORDER BY [Date]) as rownum, *
    FROM PartsInventory
) x
WHERE x.Part = PartsInventory.Part AND x.Date = PartsInventory.Date

--Recursive CTE solution:
WITH InvTotalsCTE AS
(
    SELECT [Date], RowNumWithinPart, Part, Net_Quantity, Net_Quantity as Closing_Inventory
    FROM PartsInventory
    WHERE RowNumWithinPart = 1

    UNION ALL

    SELECT Q.[Date], Q.RowNumWithinPart, Q.Part, Q.Net_Quantity, iif(P.Closing_Inventory < 0, 0, P.Closing_Inventory) + Q.Net_Quantity as Closing_Inventory
    FROM InvTotalsCTE as P
        JOIN PartsInventory Q ON (Q.RowNumWithinPart - 1) = P.RowNumWithinPart AND Q.Part = P.Part
)
SELECT [Date], Part, RowNumWithinPart, Net_Quantity, Closing_Inventory
FROM InvTotalsCTE
ORDER BY Part, RowNumWithinPart

Answer 2

如果您提供了一些示例数据会更容易，但我将如何解决这个问题：

Declare @testData Table ([Date] date, Part char(1), NetQuantity int);
 Insert Into @testData ([Date], Part, NetQuantity)
 Values ('2021-06-30', 'A', 1000)
      , ('2021-07-31', 'A', -150)
      , ('2021-08-31', 'A', -200)
      , ('2021-09-30', 'A', -500)
      , ('2021-10-31', 'A', -200)
      , ('2021-11-30', 'A', -200)
      , ('2021-12-31', 'A',   50)
      , ('2021-06-30', 'B',  100)
      , ('2021-07-31', 'B',  -80)
      , ('2021-08-31', 'B',   20)
      , ('2021-09-30', 'B',  -30)
      , ('2021-10-31', 'B',  -35)
      , ('2021-11-30', 'B',  -40)
      , ('2021-12-31', 'B', -150);

   With runningTotal
     As (
 Select *
      , Inventory = sum(td.NetQuantity) over(Partition By td.Part Order By td.[Date])
   From @testData               td      
        )
 Select [Date]
      , Part
      , NetQuantity
      , ClosingInventory = iif(lag(Inventory, 1, 0) over(Partition By Part Order By [Date]) < 0, NetQuantity, Inventory)
   From runningTotal;