Get the price of purchase at the moment of the sale - sql

Question

I'm trying to get the price of purchase at the moment of the sale. There are differents prices of purchase for the same product in my table.
The price of purchase is defined by two dates. One date for the start of the validity and one other for the end of validity : DATE_DEB_VALID and DATE_FIN_VALID.
If I want to know how many I won at the time of sale (FLIGHT_DATE), I have to know the purchase price to the same period.
My flight date must be between "DATE_DEB_VALID" and "DATE_FIN_VALID"
The two tables :
TB_DW_VAB_SALES
- ID_TEC_SALES
- TRANSACTION_NUMBER
- CARRIER_CODE
- MASTER_FLIGHT_NUMBER
- FLIGHT_NUMBER
- FLIGHT_DATE
- FLIGHT_CITY_DEP
- FLIGHT_CITY_ARR
- PRODUCT_CODE
- QUANTITY
- UNIT_SALES_PRICE
- PROMOTION_CODE
- CREW_PRICE
- COMPLEMENTARY
- SALES_TYPE
- DATE_CHGT
TB_DW_VAB_PURCHASE
- PRODUCT_CODE
- PRODUCT_CODE_GUEST
- LIB_PRODUCT
- PRICE DATE_DEB_VALID
- DATE_FIN_VALID
- DATE_CHGT
My request :
SELECT (TB_DW_VAB_PURCHASE.PRODUCT_CODE),PRICE
FROM TB_DW_VAB_PURCHASE,TB_DW_VAB_SALES
WHERE FLIGHT_DATE BETWEEN DATE_DEB_VALID AND DATE_FIN_VALID AND to_char(FLIGHT_DATE,'YYYY')= '2016' OR to_char(FLIGHT_DATE,'YYYY')= '2015'
GROUP BY TB_DW_VAB_PURCHASE.PRODUCT_CODE, PRICE
Here We have the whole request (for informations) :
SELECT to_char(DATE_VENTE,'MM/YYYY'),sum(MARGE_TOTALE) FROM (
SELECT
CTE1.CA AS CHIFFRE_AFFAIRE_TOTAL,
CTE2.PRICE AS COUT_UNITAIRE,
CTE1.FLIGHT_DATE as DATE_VENTE,
CTE1.QTE*CTE2.PRICE AS COUT_ACHAT,
(CTE1.CA-CTE1.QTE*CTE2.PRICE) AS MARGE,
sum((CTE1.CA-CTE1.QTE*CTE2.PRICE)) as MARGE_TOTALE
FROM (
SELECT PRODUCT_CODE,
sum(QUANTITY*UNIT_SALES_PRICE) AS CA,
FLIGHT_DATE,
sum(QUANTITY) as QTE
FROM TB_DW_VAB_SALES
where SALES_TYPE = 'SALES' and to_char(FLIGHT_DATE,'YYYY')= '2015' OR to_char(FLIGHT_DATE,'YYYY')= '2016'
group by to_char(FLIGHT_DATE,'MM'),FLIGHT_DATE,PRODUCT_CODE
ORDER BY to_char(FLIGHT_DATE,'MM') ASC
)CTE1
inner join
(
SELECT (TB_DW_VAB_PURCHASE.PRODUCT_CODE),PRICE
FROM TB_DW_VAB_PURCHASE,TB_DW_VAB_SALES
WHERE FLIGHT_DATE BETWEEN DATE_DEB_VALID AND DATE_FIN_VALID AND to_char(FLIGHT_DATE,'YYYY')= '2016' OR to_char(FLIGHT_DATE,'YYYY')= '2015'
GROUP BY TB_DW_VAB_PURCHASE.PRODUCT_CODE, PRICE
) CTE2
on CTE1.PRODUCT_CODE=CTE2.PRODUCT_CODE
group by to_char(FLIGHT_DATE,'MM'), FLIGHT_DATE, 'MM', CTE1.FLIGHT_DATE, (CTE1.CA-CTE1.QTE*CTE2.PRICE),
CTE1.CA, CTE1.QTE, CTE2.PRICE, CTE1.QTE*CTE2.PRICE
) group by to_char(DATE_VENTE,'MM/YYYY') ORDER BY to_char(DATE_VENTE,'MM/YYYY') ASC;
Thank you !

Answer 1

As Nemeros already pointed out, you have a cross-join in your CTE2 subquery, as you aren't linking sales and purchases together - other than by the flight/valid dates, but across all products, which can't be what you intended.
It looks like you're calculating things in your inline views (naming them CTE1/2 is slightly confusing as that usually refers to common table expressions or subquery factoring) which you then use for further calculations, but I don't think the intermediate steps or values are needed.
It (looks* like your query can be simplified to something like:
select to_char(trunc(s.flight_date, 'MM'),'MM/YYYY') as mois_du_sales,
sum(s.quantity*(s.unit_sales_price - p.price)) as marge_totale
from tb_dw_vab_sales s
join tb_dw_vab_purchase p
on p.product_code = s.product_code
and s.flight_date between p.date_deb_valid and p.date_fin_valid
where s.sales_type = 'SALES'
and s.flight_date >= date '2015-01-01'
and s.flight_date < date '2017-01-01'
group by trunc(s.flight_date, 'MM')
order by trunc(s.flight_date, 'MM') asc;
Of course without sample data or expected results I can't verify I've interpreted it correctly. It may also not solve all of your performance problems - you still need to look at the tables and indexes and the execution plan to see what it's actually doing, and you might want to consider partitioning if you have a large volume of data and a convenient partition key (like sales year).
Using a filter like to_char(FLIGHT_DATE,'YYYY')= '2015' forces every value in that column (or at least those that match sales_type, if that is indexed and selective enough) to be converted to strings and then compared to the fixed value - twice as you're checking both years separately - and that stops any index on flight_date being used. Using a date range allows the index to be used, though unless you have data spanning many years it may still not be selective enough to be used, and the optimiser may still choose full table scans. But only one of each table, not two as you were potentially doing.

Answer 2

I will just fix your initial query till you just forgotted your basic algebra (AND have more priority than OR) :
SELECT (TB_DW_VAB_PURCHASE.PRODUCT_CODE),PRICE
FROM TB_DW_VAB_PURCHASE,TB_DW_VAB_SALES
WHERE FLIGHT_DATE BETWEEN DATE_DEB_VALID AND DATE_FIN_VALID AND
(to_char(FLIGHT_DATE,'YYYY')= '2016' OR to_char(FLIGHT_DATE,'YYYY')= '2015')
GROUP BY TB_DW_VAB_PURCHASE.PRODUCT_CODE, PRICE
Now if you had used normalized join, you would have also seen that you had forgotten a join clause on the product_code, so finally :
SELECT PUR.PRODUCT_CODE, PRICE
FROM TB_DW_VAB_PURCHASE PUR
inner join TB_DW_VAB_SALES SAL ON PUR.PRODUCT_CODE = PUR.PRODUCT_CODE
AND FLIGHT_DATE BETWEEN DATE_DEB_VALID AND DATE_FIN_VALID
WHERE to_char(FLIGHT_DATE,'YYYY')= '2016' OR to_char(FLIGHT_DATE,'YYYY')= '2015'
GROUP BY PUR.PRODUCT_CODE, PRICE