How to vertically flip data in BQL? - sql

Question

Title. For example, I have below data:
Key1 Key2 Cost Qty_LIFO Date
Red A 2 19 1/4/2018
Red A 3 18 1/3/2018
Red C 4 7 1/2/2018
Red A 5 16 1/1/2018
Blu B 21 91 1/4/2018
Blu B 31 81 1/3/2018
Blu D 41 70 1/2/2018
Blu D 51 60 1/1/2018
The goal is to transform the data to look like below. Flip the quantity column, while also taking into account the Keys/categories
Key1 Key2 Cost Qty_FIFO Date
Red A 2 16 1/4/2018
Red A 3 18 1/3/2018
Red C 4 7 1/2/2018
Red A 5 19 1/1/2018
Blu B 21 81 1/4/2018
Blu B 31 91 1/3/2018
Blu D 41 60 1/2/2018
Blu D 51 70 1/1/2018
or like this (Qty_FIFO is flipped and added to the first example at the top):
Key1 Key2 Cost Qty_LIFO Qty_FIFO Date
Red A 2 19 16 1/4/2018
Red A 3 18 18 1/3/2018
Red C 4 7 7 1/2/2018
Red A 5 16 19 1/1/2018
Blu B 21 91 81 1/4/2018
Blu B 31 81 91 1/3/2018
Blu D 41 70 60 1/2/2018
Blu D 51 60 70 1/1/2018
The purpose of this is to calculate LIFO and FIFO costs.
I need to take Qty_LIFO column (which is sorted by date, descending), flip it vertically (so the data becomes Date ASC), and re-add it to the table without changing the sorting of the Costs column.
Basically, I need to pair the newest Cost data with the oldest Qty data and continue from there.

Answer 1

This is a hack, which only works if you have access to row_number()
CREATE TABLE existing_qry(
Key1 VARCHAR(3) NOT NULL
,Key2 VARCHAR(1) NOT NULL
,Cost INTEGER NOT NULL
,Qty_LIFO INTEGER NOT NULL
,Date DATE NOT NULL
);
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Red','A',2,19,'1/4/2018');
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Red','A',3,18,'1/3/2018');
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Red','C',4,7,'1/2/2018');
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Red','A',5,16,'1/1/2018');
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Blu','B',21,91,'1/4/2018');
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Blu','B',31,81,'1/3/2018');
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Blu','D',41,70,'1/2/2018');
INSERT INTO existing_qry(Key1,Key2,Cost,Qty_LIFO,Date) VALUES ('Blu','D',51,60,'1/1/2018');
with cte as (
select
*
, row_number() over(partition by key1 order by date ASC) rn_asc
, row_number() over(partition by key1 order by date DESC) rn_desc
from existing_qry
)
select
t.Key1, t.Key2, t.Cost, t.Qty_LIFO, flip.Qty_LIFO as Qty_FIFO, t.Date, t.rn_asc, t.rn_desc
from cte as t
inner join cte as flip on t.key1 = flip.key1 and t.rn_asc = flip.rn_desc
It calculates 2 numbers for each row in opposite date order, then aligns rows by requiring these to be equal through a self join. This has the impact of reversing the LIFO numbers (or "flipping" that column).
Key1 Key2 Cost Qty_LIFO Qty_FIFO Date rn_asc rn_desc
---- ------ ------ ------ ---------- ---------- --------------------- -------- ---------
1 Blu B 21 91 60 04.01.2018 00:00:00 4 1
2 Blu B 31 81 70 03.01.2018 00:00:00 3 2
3 Blu D 41 70 81 02.01.2018 00:00:00 2 3
4 Blu D 51 60 91 01.01.2018 00:00:00 1 4
5 Red A 2 19 16 04.01.2018 00:00:00 4 1
6 Red A 3 18 7 03.01.2018 00:00:00 3 2
7 Red C 4 7 18 02.01.2018 00:00:00 2 3
8 Red A 5 16 19 01.01.2018 00:00:00 1 4
https://rextester.com/LQBVD29253