How to use fiscal calendar in a query - sql

Question

I have two tables "Fiscal_calendar" and "Sales". The 2 tables do not link together. I want to be able to write a query that calculates the total sales made a in a week according to the fiscal calendar. Is this possible? Our fiscal calendar starts from December 1st and every month end falls on a friday.
Any help would be appreciated.
Fiscal_calendar
Period Period1_StartDate Period1_EndDate Period2_StartDate Period2_EndDate...........
2018 01/12/2017 29/12/2017 30/01/2018 26/01/2018
Sales
Sales_order_no Amount Date Customer
111 20453 03/12/2017 abc
112 23154 04/12/2017 bbb
113 20201 10/12/2017 ddd
114 39012 11/12/2017 ccc
115 11111 18/12/2017 eee
116 22222 25/12/2017 uuu
So there are 4 weeks between Period 1 startdate and enddate. And the first 2 sales fall under week 1. So total sales for Week 1 would be 43607
OUTPUT
WEEK Total_Sales
W1 43607
W2 59213
W3 11111
W4 22222

Answer 1

Two uses of CROSS APPLY to fix (normalise) the fiscal calendar table in to something useful, then a simple GROUP BY.
The WHERE clause picks the row from the fiscal calendar, that the period or week (or whatever) from the derived views.
The LEFT JOIN is in case of no sales in that week.
The ON clause looks at the fiscal calendar AND the week derived view, just in case the derived view describes a week that's not really in that year.
SELECT
p.year_id,
p.period_id,
w.week_id,
SUM(s.amount) AS total_amount
FROM
fiscal_calendar c
CROSS APPLY
(
SELECT period, 1, period1_startDate, period1_endDate
UNION ALL SELECT period, 2, period2_startDate, period2_endDate
..
UNION ALL SELECT period, 13, period13_startDate, period13_endDate
)
AS p(year_id, period_id, startDate, endDate)
CROSS APPLY
(
SELECT 1, DATEADD(d, 0, startDate), DATEADD(d, 6, startDate)
UNION ALL SELECT 2, DATEADD(d, 7, startDate), DATEADD(d, 13, startDate)
UNION ALL SELECT 3, DATEADD(d, 14, startDate), DATEADD(d, 20, startDate)
UNION ALL SELECT 4, DATEADD(d, 21, startDate), DATEADD(d, 27, startDate)
)
AS w(week_id, startDate, endDate)
LEFT JOIN
sales s
ON s.date BETWEEN c.period1_startdate AND c.period13_enddate
AND s.date BETWEEN w.startDate AND w.endDate
WHERE
p.year_id = 2018
AND p.period_id = 1
GROUP BY
p.year_id,
p.period_id,
w.week_id

Answer 2

UNPIVOT is a useful TSQL operator to use when normalizing data. Basically it takes a row and makes new rows - one for each column specified in a list of columns.
In the case below I unpivot on the PeriodX_StartDate columns. I originally did two unpivots, the other on PeriodX_EndDate. The end date was unnecessary, but in case you're wondering how I joined the two unpivots: I used a cross apply to generate a key for each week and had a predicate that tested if they were equal. Without the predicate you get a Cartesian product.
To generate the weeks I used a CROSS APPLY that is very similar to the one in #MatBailie's answer, except I avoided the unions by generating a set of arguments for each week that are passed into functions that generate the dates.
WITH normalizedCal AS (
SELECT [Period], MonthStart, WeekKey, StartDate, CutoffDate
FROM fiscal_calendar cal
UNPIVOT (
MonthStart FOR StartKey IN (Period1_StartDate, Period2_StartDate, ..., PeriodN_StartDate)
) AS startInfo
CROSS APPLY (
SELECT 'WK' + CAST((StartIndex / 7) + 1 AS char(1)) [WeekKey]
, DATEADD(day, startIndex, MonthStart) [StartDate]
, DATEADD(day, EndIndex, MonthStart) [CutoffDate]
FROM (
VALUES ( 0, 7 ), ( 7, 14 ), ( 14, 21 ), ( 21, 28 )
) rangeValues ( StartIndex, EndIndex )
) weekInfo
)
SELECT [Period], MonthStart, WeekKey, COALESCE(SUM(AMOUNT), 0) [Total_Sales]
FROM normalizedCal nc
LEFT JOIN sales ON sales.Date >= nc.StartDate AND sales.Date < nc.CutoffDate
GROUP BY [Period], MonthStart, WeekKey
ORDER BY [Period], MonthStart, WeekKey
Working Example with sample data:
WITH normalizedCal AS (
SELECT [Period], MonthStart, WeekKey, StartDate, CutoffDate
FROM (
VALUES (2018, '2017-12-01', '2017-12-29', '2017-12-30', '2018-01-26')
) cal ([Period], Period1_StartDate, Period1_EndDate, Period2_StartDate, Period2_EndDate)
UNPIVOT (
MonthStart FOR StartKey IN (Period1_StartDate, Period2_StartDate)
) AS startInfo
CROSS APPLY (
SELECT 'WK' + CAST((StartIndex / 7) + 1 AS char(1)) [WeekKey]
, DATEADD(day, startIndex, MonthStart) [StartDate]
, DATEADD(day, EndIndex, MonthStart) [CutoffDate]
FROM (
VALUES ( 0, 7 ), ( 7, 14 ), ( 14, 21 ), ( 21, 28 )
) rangeValues ( StartIndex, EndIndex )
) weekInfo
),
sales AS (
SELECT [Sales_order_no],[Amount], CAST([Date] as DATE) [Date],[Customer]
FROM (VALUES
(111, 20453, '2017-12-03', N'abc'),
(112, 23154, '2017-12-04', N'bbb'),
(113, 20201, '2017-12-10', N'ddd'),
(114, 39012, '2017-12-11', N'ccc'),
(115, 11111, '2017-12-18', N'eee'),
(116, 22222, '2017-12-25', N'uuu')
) [salessrc] ( [Sales_order_no],[Amount],[Date],[Customer])
)
SELECT [Period], MonthStart, WeekKey, COALESCE(SUM(AMOUNT), 0) [Total_Sales]
FROM normalizedCal nc
LEFT JOIN sales ON sales.Date >= nc.StartDate AND sales.Date < nc.CutoffDate
GROUP BY [Period], MonthStart, WeekKey
ORDER BY [Period], MonthStart, WeekKey

Answer 3

according to your provided example, You could do something like this :
SELECT *
FROM (
SELECT
[YEAR] = YEAR([Date]),
[MONTH] = MONTH([Date]),
[WEEK] = DATEPART(week, [Date]) - DATEPART(week, DATEADD(MONTH, DATEDIFF(MONTH, 0, [Date]), 0)),
SUM(Amount) OVER(PARTITION BY YEAR([Date]), DATEPART(week, [Date]) - DATEPART(week, DATEADD(MONTH, DATEDIFF(MONTH, 0, [Date]), 0))+ 1) AS TotalSales
FROM Sales
)D
GROUP BY
[YEAR],
[MONTH],
[WEEK],
TotalSales
this will give you the same results that you've provided in your example, however, since we don't know how your Fiscal_calendar is configured you will need to see is it a mandatory to link it with it or not.