How to write SQL for stock quantity that requires calculation from previous orders - sql

Question

I have two tables, one for current total stock of products and one for the product orders.
STOCK_TB
PRODUCT_ID STOCK_QTY
A 20
B 15
C 10
ORDER_TB
ORDER_DATE PRODUCT_ID ORDER QTY
2015-03-01 A 5
2015-03-02 A 3
2015-03-02 B 4
2015-03-03 C 1
2015-03-04 C 3
I'd like to select data for a monthly-stock quantity report that looks like this. Assume the report was built on March 5th
Stock Quantity of March:
Daily Stock Qty
Product ID 1 2 3 4 5 6 7 ... 28 29 30 31
A 23 20 20 20 20 0 0 0 0 0 0
B 19 15 15 15 15 0 0 0 0 0 0
C 14 14 13 10 10 0 0 0 0 0 0
The stock quantity for previous dates is based on the closing day (I.E: March 2nd above refers to March 2nd 23:59:99.999)
Any dates that goes beyond the current date will have a quantity of 0
We don't have a table for keeping daily-stocks, just the current stock. So this means for getting stocks of previous dates, I'd have to add the amount of product orders backwards.
How do you write this type of query? For the date columns, I can have them fixed from 1 to 31, since I can just hide the unused dates based on the month in my application. But I'm not really sure how I can write logic in SQL for adding order quantity to the current stock on previous dates.

Answer 1

Query example for 6 days (the other 25 days are the same :-)
DECLARE #FirstOfMonth AS DATETIME = DATEADD(month, DATEDIFF(month, 0, GETDATE()), 0)
SELECT
CASE WHEN DAY(GETDATE()) < 1 THEN 0 ELSE S.STOCK_QTY + ISNULL((SELECT SUM(O.ORDER_QTY) FROM ORDER_TB O WHERE O.PRODUCT_ID = S.PRODUCT_ID AND O.ORDER_DATE > DATEADD(day, 0, #FirstOfMonth)), 0) END _1,
CASE WHEN DAY(GETDATE()) < 2 THEN 0 ELSE S.STOCK_QTY + ISNULL((SELECT SUM(O.ORDER_QTY) FROM ORDER_TB O WHERE O.PRODUCT_ID = S.PRODUCT_ID AND O.ORDER_DATE > DATEADD(day, 1, #FirstOfMonth)), 0) END _2,
CASE WHEN DAY(GETDATE()) < 3 THEN 0 ELSE S.STOCK_QTY + ISNULL((SELECT SUM(O.ORDER_QTY) FROM ORDER_TB O WHERE O.PRODUCT_ID = S.PRODUCT_ID AND O.ORDER_DATE > DATEADD(day, 2, #FirstOfMonth)), 0) END _3,
CASE WHEN DAY(GETDATE()) < 4 THEN 0 ELSE S.STOCK_QTY + ISNULL((SELECT SUM(O.ORDER_QTY) FROM ORDER_TB O WHERE O.PRODUCT_ID = S.PRODUCT_ID AND O.ORDER_DATE > DATEADD(day, 3, #FirstOfMonth)), 0) END _4,
CASE WHEN DAY(GETDATE()) < 5 THEN 0 ELSE S.STOCK_QTY + ISNULL((SELECT SUM(O.ORDER_QTY) FROM ORDER_TB O WHERE O.PRODUCT_ID = S.PRODUCT_ID AND O.ORDER_DATE > DATEADD(day, 4, #FirstOfMonth)), 0) END _5,
CASE WHEN DAY(GETDATE()) < 6 THEN 0 ELSE S.STOCK_QTY + ISNULL((SELECT SUM(O.ORDER_QTY) FROM ORDER_TB O WHERE O.PRODUCT_ID = S.PRODUCT_ID AND O.ORDER_DATE > DATEADD(day, 5, #FirstOfMonth)), 0) END _6
FROM STOCK_TB S
Note that I've used > DATEADD instead of >= DATEADD but I'm not so sure... The order you put the first of the month when are counted?
Second solution, but I don't think the complexity will change very much:
DECLARE #FirstOfMonth AS DATETIME = DATEADD(month, DATEDIFF(month, 0, GETDATE()), 0)
DECLARE #Today AS DATETIME = DATEADD(dd, DATEDIFF(dd, 0, GETDATE()), 0)
;WITH Days(d, dat) AS
(
SELECT 1, #FirstOfMonth
UNION ALL
SELECT d+1, DATEADD(day, 1, dat) FROM Days WHERE d < DATEPART(day, #today)
)
, Work1 AS (
SELECT PRODUCT_ID, STOCK_QTY + ISNULL((SELECT SUM(O.ORDER_QTY) FROM ORDER_TB O WHERE O.PRODUCT_ID = S.PRODUCT_ID AND O.ORDER_DATE > dat), 0) STOCK_TB, d FROM STOCK_TB S, Days
)
SELECT PRODUCT_ID,
ISNULL(MAX(CASE WHEN d = 1 THEN STOCK_TB END), 0) _1,
ISNULL(MAX(CASE WHEN d = 2 THEN STOCK_TB END), 0) _2,
ISNULL(MAX(CASE WHEN d = 3 THEN STOCK_TB END), 0) _3,
ISNULL(MAX(CASE WHEN d = 4 THEN STOCK_TB END), 0) _4,
ISNULL(MAX(CASE WHEN d = 5 THEN STOCK_TB END), 0) _5,
ISNULL(MAX(CASE WHEN d = 6 THEN STOCK_TB END), 0) _6,
ISNULL(MAX(CASE WHEN d = 7 THEN STOCK_TB END), 0) _7,
ISNULL(MAX(CASE WHEN d = 8 THEN STOCK_TB END), 0) _8
FROM Work1 GROUP BY PRODUCT_ID
Here I use a fancy recursive query to build a table of days 1...(today), then I build a Work1 intermediate that has all the stock quantities day by day (so x products * y days rows), and then I group them
Third possibility: double recursive query (one to calculate the numbers 1...31 and one to do a running total), plus the final GROUP BY nearly identical to the previous example.
DECLARE #FirstOfMonth AS DATETIME = DATEADD(month, DATEDIFF(month, 0, GETDATE()), 0)
DECLARE #Today AS DATETIME = DATEADD(dd, DATEDIFF(dd, 0, GETDATE()), 0)
;WITH Days(d, dat) AS
(
SELECT DATEPART(day, #Today), #Today dat
UNION ALL
SELECT d-1, DATEADD(day, -1, dat) dat
FROM Days
WHERE d > 1
)
# Product Days x STOCK_TB with a LEFT JOIN on ORDER_TB.
, Work1 AS (
SELECT S.PRODUCT_ID, d, dat, S.STOCK_QTY, ISNULL(O.ORDER_QTY, 0) ORDER_QTY
FROM Days
CROSS JOIN STOCK_TB S # Full cartesian product, JOIN without conditions
LEFT JOIN ORDER_TB O ON dat = O.ORDER_DATE AND S.PRODUCT_ID = O.PRODUCT_ID
)
# Second recursive query to do the running total
, Days2(PRODUCT_ID, d, dat, STOCK_QTY) AS
(
SELECT PRODUCT_ID, d, dat, STOCK_QTY
FROM Work1
WHERE d = DATEPART(day, #Today)
UNION ALL
SELECT d.PRODUCT_ID, d.d - 1, w.dat, d.STOCK_QTY + w.ORDER_QTY
FROM Days2 d
INNER JOIN Work1 w ON d.PRODUCT_ID = w.PRODUCT_ID AND d.d /* - 1 */ = w.d
WHERE d.d > 1
)
SELECT PRODUCT_ID,
ISNULL(MAX(CASE WHEN d = 1 THEN STOCK_QTY END), 0) _1,
ISNULL(MAX(CASE WHEN d = 2 THEN STOCK_QTY END), 0) _2,
ISNULL(MAX(CASE WHEN d = 3 THEN STOCK_QTY END), 0) _3,
ISNULL(MAX(CASE WHEN d = 4 THEN STOCK_QTY END), 0) _4,
ISNULL(MAX(CASE WHEN d = 5 THEN STOCK_QTY END), 0) _5,
ISNULL(MAX(CASE WHEN d = 6 THEN STOCK_QTY END), 0) _6,
ISNULL(MAX(CASE WHEN d = 7 THEN STOCK_QTY END), 0) _7,
ISNULL(MAX(CASE WHEN d = 8 THEN STOCK_QTY END), 0) _8
FROM Days2 GROUP BY PRODUCT_ID
Note the /* - 1 */ commented part. Uncommenting it you control how the value of the first of the month is used.