Calculating closing (sums of two views, including the past) - sql

Question

I am trying to get the remaining number of working units for each month, of a sum between a bought number of working unit, and a consumed number of working unit.
I tried two possibilities, but both have flaws :
In the first test, I created a "Months" table that contains every month and every year, in order to show all months in the final matrix I wish to create with these data. With this one, I get the closing whenever there is a consumed working unit, but when there is not, the column is "empty", because it does not get the last closing.
USE OTRS_Revised
SELECT [Customer], CASE WHEN [Year] < 2016 THEN 1 ELSE [Year] END AS [Year], CASE WHEN [Year] < 2016 THEN 0 ELSE [Month] END AS [Month], [Closing] AS Total, SUM([Closing])
OVER (PARTITION BY [Customer] ORDER BY [Year], [Month] ROWS UNBOUNDED PRECEDING) AS Closing
FROM [dbo].[WU_Closing_View]
WHERE [Customer] IN ('CustomerList')
GROUP BY [Customer], [Year], [Month], [Closing]
UNION ALL
SELECT '' AS Customer, CASE WHEN [Year] < 2016 THEN 1 ELSE [Year] END AS [Year], CASE WHEN [Year] < 2016 THEN 0 ELSE [Month] END AS [Month], '' AS Total, '' AS Sum_bought
FROM [dbo].Months
WHERE [Year] <= 2016
GROUP BY Year, Month
ORDER BY Customer, Year, Month
I also tried to do it "month by month", with the below query. It works for one month, but I can't find any way to use this to get the results for each month of the year 2016.
SELECT
(SELECT SUM(Closing) AS Expr1
FROM OTRS_Revised.dbo.WU_Bought_View
WHERE (Customer LIKE 'SomeCustomer') AND (DATEADD(Year, Year - 1900, DATEADD(Month, Month - 1, DATEADD(day, 0, 0))) <= DATEADD(Year, 2016 - 1900, DATEADD(Month, 5 - 1, DATEADD(day, 0, 0))))
GROUP BY Customer)
+
(SELECT SUM(Closing) AS Expr1
FROM OTRS_Revised.dbo.WU_Consumed_View
WHERE (Customer LIKE 'SomeCustomer') AND (DATEADD(Year, Year - 1900, DATEADD(Month, Month - 1, DATEADD(day, 0, 0))) <= DATEADD(Year, 2016 - 1900, DATEADD(Month, 5 - 1, DATEADD(day, 0, 0))))
GROUP BY Customer) AS Expr1,
[Month]
FROM OTRS_Revised.dbo.Months
GROUP BY [Month]

Answer 1

SELECT b.*
FROM
( SELECT CASE WHEN [Year] < 2016 THEN 1 ELSE [Year] END AS [Year], CASE WHEN [Year] < 2016 THEN 0 ELSE [Month] END AS [Month]
FROM [dbo].Months
WHERE [Year] <= 2016
GROUP BY Year, Month
ORDER BY Customer, Year, Month ) AS a
LEFT OUTER JOIN
(SELECT [Customer], CASE WHEN [Year] < 2016 THEN 1 ELSE [Year] END AS [Year], CASE WHEN [Year] < 2016 THEN 0 ELSE [Month] END AS [Month], [Closing] AS Total, SUM([Closing])
OVER (PARTITION BY [Customer] ORDER BY [Year], [Month] ROWS UNBOUNDED PRECEDING) AS Closing
FROM [dbo].[WU_Closing_View]
WHERE [Customer] IN ('CustomerList')
GROUP BY [Customer], [Year], [Month], [Closing]) AS b
ON a.Month = b.Month )
In your approach when you do union the rows that don't have matching months is getting removed. Since you want the months that does not have closing match as well, you need to use the left outer join

Answer 2

Something like this perhaps
DECLARE #T TABLE (ID INT, ProductID INT, TrDate DATE,InOut VARCHAR(10),Amount INT)
INSERT INTO #T VALUES
(1 ,1, '2016-01-01', 'I', 100),
(2 ,2, '2016-01-01', 'I', 100),
(3 ,3, '2016-02-01', 'I', 100),
(4 ,4, '2016-03-01', 'I', 100),
(5 ,1, '2016-03-01', 'I', 100),
(6 ,2, '2016-04-01', 'O', 10),
(7 ,3, '2016-05-01', 'I', 100),
(8 ,5, '2016-05-01', 'I', 100),
(9 ,5, '2016-05-01', 'O', 100),
(10 ,6, '2016-05-01', 'I', 100)
declare #m table (id int, menddate date)
insert #m values
(1,'2015-12-31'),(2,'2016-01-31'),(3,'2016-02-29'),(4,'2016-03-31'),
(5,'2016-04-30'),(6,'2016-05-31'),(7,'2016-06-30'),(4,'2016-07-31')
Select *
from
(
select -- t.*
x.xproductid , x.xyyyymm,
SUM(t.total) OVER (partition by x.xproductid
ORDER BY x.xyyyymm
ROWS UNBOUNDED PRECEDING) AS CumulativeTotal
from
(
SELECT t.ProductID tproductid, year(t.trdate) * 100 + month(t.trdate) tyyyymm,
sum(case when t.Inout = 'I' then t.Amount else t.amount * -1 end) as total
FROM #T t
group by ProductID, year(t.trdate) * 100 + month(t.trdate)
) t
right outer join
(select distinct productid as xproductid,year(m.menddate) * 100 + month(m.menddate) xyyyymm from #t t, #m m) x on x.xproductid = t.tproductid and x.xyyyymm = t.tyyyymm
) z
where z.xyyyymm >= 201601
order by z.xProductID,z.xyyyymm
Note the use of a right outer join to get all the month ends for all products