Embedded Case statement - sql

Question

I have a query (which works) however, I want to be able to make it so I don't need to rerun the query to change the date ranges.
What I have here is it is pulling customer name and the amount and summing it. I need it broken down monthly by fiscal year, so what I would just do is change the year and rerun it, but I want to make the query more efficient so it doesn't need to be changed in the future (except the end date for a new FY) So after it takes the sum of each month (July to June), I want it to show the FY. So the date ranges from 2014 to 2015 should show a column called FY and have 2015 in the column. I am pretty certain I am on the right track but not there. I only showed what I was working on for the first case (which will be replicated for the other cases)
Could this just be an IF statement loop instead of another case statement? Any help with direction would be great to help me learn. I got this far, but just want to make the code more efficient for someone to use
Declare #StartDate datetime
Declare #EndDate datetime
set #StartDate = convert(datetime, '07/01/2014')
set #EndDate = convert(datetime, '06/30/2017')
select
d.CUSTNMBR Customer_ID,
d.CUSTNAME Customer_Name,
'A' as Company,
sum(case when month(d.DOCDATE) = 7
then d.SALES else 0 end) July,
case when datetime between '07/01/2016' and '06/30/2017' as '2017'
when datetime between '07/01/2015' and '06/30/2016' as '2016'
when datetime between '07/01/2014' and '06/30/2015' as '2015'
END
sum(case when month(d.DOCDATE) = 8
then d.SALES else 0 end) August,
sum(case when month(d.DOCDATE) = 9
then d.SALES else 0 end) September,
sum(case when month(d.DOCDATE) = 10
then d.SALES else 0 end) October,
sum(case when month(d.DOCDATE) = 11
then d.SALES else 0 end) November,
sum(case when month(d.DOCDATE) = 12
then d.SALES else 0 end) December,
sum(case when month(d.DOCDATE) = 1
then d.SALES else 0 end) January,
sum(case when month(d.DOCDATE) = 2
then d.SALES else 0 end) February,
sum(case when month(d.DOCDATE) = 3
then d.SALES else 0 end) March,
sum(case when month(d.DOCDATE) = 4
then d.SALES else 0 end) April,
sum(case when month(d.DOCDATE) = 5
then d.SALES else 0 end) May,
sum(case when month(d.DOCDATE) = 6
then d.SALES else 0 end) June,
sum(d.SALES) 'Year to Date'
from
(select s.DOCDATE, s.CUSTNMBR, c.CUSTNAME,
case s.SOPTYPE
when 3 then s.DOCAMNT
when 4 then s.DOCAMNT*-1
end SALES
from a1.dbo.S200 s
left outer join a1.dbo.RM1 c
on s.CUSTNMBR = c.CUSTNMBR
where s.VOIDSTTS = 0
and s.SOPTYPE in (3,4) and
s.DOCDATE between #StartDate and #EndDate -- Fiscal Year
) d
group by d.CUSTNMBR, d.CUSTNAME

Answer 1

Consider a recursive CTE that iterates through each fiscal year and then run a CROSS JOIN...WHERE (CROSS APPLY can work too) on your existing derived table. The WHERE conditions DOCDate to fiscal year start and end ranges. And going forward, simply change the declaration lines for each fiscal year. In fact, you can set up #end INT = 2020 to anticipate future data!
DECLARE #start INT = 2014;
DECLARE #end INT = 2016;
WITH fiscalyears AS (
SELECT #start + 1 AS FY,
CONVERT(datetime, cast(1 as varchar)+'/'+cast(7 as varchar)+'/'+cast(#start as varchar), 103) As StartRng,
CONVERT(datetime, cast(30 as varchar)+'/'+cast(6 as varchar)+'/'+cast(#start + 1 as varchar), 103) As EndRng
UNION ALL
SELECT FY + 1,
CONVERT(datetime, cast(1 as varchar)+'/'+cast(7 as varchar)+'/'+cast(FY + 1 as varchar), 103),
CONVERT(datetime, cast(30 as varchar)+'/'+cast(6 as varchar)+'/'+cast(FY + 2 as varchar), 103)
FROM fiscalyears
WHERE FY < #end
)
-- FY StartRng EndRng
-- 2015 2014-07-01 00:00:00.000 2015-06-30 00:00:00.000
-- 2016 2015-07-01 00:00:00.000 2016-06-30 00:00:00.000
---2017 2016-07-01 00:00:00.000 2017-06-30 00:00:00.000
select
d.CUSTNMBR Customer_ID,
d.CUSTNAME Customer_Name,
'A' as Company,
d.FY As FY,
sum(case when month(d.DOCDATE) = 7 then d.SALES else 0 end) AS July,
sum(case when month(d.DOCDATE) = 8 then d.SALES else 0 end) AS August,
sum(case when month(d.DOCDATE) = 9 then d.SALES else 0 end) AS September,
sum(case when month(d.DOCDATE) = 10 then d.SALES else 0 end) AS October,
sum(case when month(d.DOCDATE) = 11 then d.SALES else 0 end) AS November,
sum(case when month(d.DOCDATE) = 12 then d.SALES else 0 end) AS December,
sum(case when month(d.DOCDATE) = 1 then d.SALES else 0 end) AS January,
sum(case when month(d.DOCDATE) = 2 then d.SALES else 0 end) AS February,
sum(case when month(d.DOCDATE) = 3 then d.SALES else 0 end) AS March,
sum(case when month(d.DOCDATE) = 4 then d.SALES else 0 end) AS April,
sum(case when month(d.DOCDATE) = 5 then d.SALES else 0 end) AS May,
sum(case when month(d.DOCDATE) = 6 then d.SALES else 0 end) AS June,
sum(d.SALES) 'Year to Date'
from
(select f.FY, s.DOCDATE, s.CUSTNMBR, c.CUSTNAME,
case s.SOPTYPE
when 3 then s.DOCAMNT
when 4 then s.DOCAMNT*-1
end SALES
from a1.dbo.S200 s
left outer join a1.dbo.RM1 c
on s.CUSTNMBR = c.CUSTNMBR
cross join fiscalyears f
where s.VOIDSTTS = 0
and s.SOPTYPE in (3,4)
and s.DOCDATE between f.StartRng and f.EndRng) d
group by d.CUSTNMBR,
d.CUSTNAME,
d.FY

Answer 2

Do you have access to create functions? If so, create a (re-usable) scalar function for FY. E.g.:
CREATE FUNCTION dbo.FY (input_date as smalldatetime)
RETURNS int
BEGIN
DECLARE #return int
SET #return = year(input_date)
IF month(input_date) <= 6
#return = #return + 1
RETURN #return
END
Then Use the function in your SELECT...
SELECT
...,
FY(d.DOCDATE),
...
FROM
...
Otherwise, you can inline the function -- e.g.:
SELECT
...,
CASE WHEN Month(d.DOCDATE) <= 6 THEN 1 ELSE 0 END + year(d.DOCDATE) as FY
...
Also worth mentioning -- it looks like you're naming your Months the hard way.
Try this instead:
DATENAME(month,d.DOCDATE),
Also...
Suggest keeping your query's data and layout separated a bit more (for example, where you're create a horizontal layout for your months) -- you might like what PIVOT (https://technet.microsoft.com/en-us/library/ms177410(v=sql.105).aspx) can do for you...
If you do this, you can keep your query's data a little cleaner -- and then use it's output in other queries as well. More re-usability = less work for you! :-)

Answer 3

I am proposing that the S200 table shall have computed columns as helper like this:
create table S200
(
CUSTNMBR varchar(10) null,
DOCDATE date null,
SOPTYPE int null,
DOCAMNT money null,
FISCALYEAR as datepart(year, DOCDATE) +
case when datepart(month, DOCDATE) >= 7 then 1 else 0 end,
FISCALMONTHNAME as (datename(month, DOCDATE))
)
Then we can simply PIVOT it like:
select
*,
isnull([January], 0) +
isnull([February], 0) +
isnull([March], 0) +
isnull([April], 0) +
isnull([May], 0) +
isnull([June], 0) +
isnull([July], 0) +
isnull([August], 0) +
isnull([September], 0) +
isnull([October], 0) +
isnull([November], 0) +
isnull([December], 0) as YearToDate
from
(
select
s.CUSTNMBR,
s.FISCALYEAR,
FISCALMONTHNAME,
case s.SOPTYPE
when 3 then s.DOCAMNT
when 4 then -s.DOCAMNT
end as SALES
from
S200 as s
)
as d
pivot
(
sum
(
d.SALES
)
for d.FISCALMONTHNAME in
(
[January],
[February],
[March],
[April],
[May],
[June],
[July],
[August],
[September],
[October],
[November],
[December]
)
)
as p
order by 2, 1
If you have a Tally table, you can mock populate S200 for 3 years starts from 2014 for 3 customers:
truncate table S200
insert
S200
select
101 + cast(3 * rand(checksum(newid())) as int),
dateadd(day, cast(3 * 365 * rand(checksum(newid())) as int),
datefromparts(2014, 1, 1)),
case when rand(checksum(newid())) > .6 then 4 else 3 end,
1 + cast(100 * rand(checksum(newid())) as int)
from
Tally
where
Id <= 1000
You may have result like this:
| CUSTNMBR | FISCALYEAR | January | February | March | April | May | June | July | August | September | October | November | December | YearToDate |
|----------|------------|---------|----------|---------|--------|---------|---------|---------|---------|-----------|---------|----------|----------|------------|
| 101 | 2014 | 160.00 | 143.00 | 139.00 | 28.00 | 247.00 | 28.00 | NULL | NULL | NULL | NULL | NULL | NULL | 745.00 |
| 102 | 2014 | 24.00 | 158.00 | -244.00 | -85.00 | 103.00 | 20.00 | NULL | NULL | NULL | NULL | NULL | NULL | -24.00 |
| 103 | 2014 | 123.00 | -167.00 | -6.00 | -30.00 | -53.00 | 15.00 | NULL | NULL | NULL | NULL | NULL | NULL | -118.00 |
| 101 | 2015 | 14.00 | 200.00 | 304.00 | 137.00 | 236.00 | -245.00 | 289.00 | 258.00 | 192.00 | -13.00 | 275.00 | 9.00 | 1656.00 |
| 102 | 2015 | -40.00 | 115.00 | -36.00 | 142.00 | -18.00 | 48.00 | -262.00 | -71.00 | 24.00 | 189.00 | -166.00 | -49.00 | -124.00 |
| 103 | 2015 | 265.00 | -103.00 | 73.00 | -50.00 | 61.00 | -268.00 | -20.00 | -214.00 | -21.00 | 50.00 | -48.00 | -204.00 | -479.00 |
| 101 | 2016 | 207.00 | -201.00 | 34.00 | 397.00 | -156.00 | 267.00 | -272.00 | -156.00 | -61.00 | -98.00 | 575.00 | 107.00 | 643.00 |
| 102 | 2016 | 27.00 | 20.00 | 199.00 | 328.00 | 44.00 | 55.00 | 447.00 | 188.00 | -57.00 | -118.00 | -1.00 | 74.00 | 1206.00 |
| 103 | 2016 | -114.00 | 333.00 | 299.00 | 145.00 | 80.00 | -8.00 | 58.00 | -151.00 | 44.00 | 799.00 | 218.00 | 14.00 | 1717.00 |
| 101 | 2017 | NULL | NULL | NULL | NULL | NULL | NULL | -154.00 | -56.00 | 192.00 | -283.00 | 371.00 | 155.00 | 225.00 |
| 102 | 2017 | NULL | NULL | NULL | NULL | NULL | NULL | 11.00 | -50.00 | 82.00 | 8.00 | 562.00 | -26.00 | 587.00 |
| 103 | 2017 | NULL | NULL | NULL | NULL | NULL | NULL | 294.00 | -93.00 | -51.00 | 29.00 | 322.00 | 392.00 | 893.00 |
Perhaps with certain indexes on those computed columns, it would give even better performance.