How to count lines backwards from log in TSQL? - sql-server

Question

I would like to count the amount of pallets on a location where you can't see on the location how many pallets there are. The only way to find out is to assume that you can count backwards from the log where it's logged what pallets have been moved to that location.
Location quantity
LOCATION QUANTITY
Loc_1 20
Log
MOVED_TO_LOCATION QUANTITY PALLET_NUMBER
Loc_1 5 13
Loc_1 5 12
Loc_1 5 11
Loc_1 5 10 <-- Count the lines from here and up (count = 4).
Loc_1 5 9
...
What I tried so far was some CROSS APPLY where I joined the location to the log and counted the results. Which returned 13 (instead of 4) because it didn't count away the ones that were already gone. So how would I count backwards and only gets the first 4 rows?
-- Query returns 13 rows.
CROSS APPLY (
SELECT PALL_NUM
FROM LOG
WHERE MOVED_TO_LOCATION = LOCATION
ORDER BY TIMESTAMP DESC) LOG
I assume you could somehow incrementally count up the amount in the log from latest to earliest timestamp until the quantity is more or equal in the log to the current quantity on the location. But I'm not sure how to do that exactly.
Pseudocode
Quantity on location = 20
Row 1 in log: 5.
20 - 5 = 15 remaining.
Row 2 in log: 5.
15 - 5 = 10 remaining.
Row 3 in log: 5.
10 - 5 = 5 remaining.
Row 4 in log: 5.
5 - 5 = 0 remaining.
Then stop and return the rows.
Edit 1: Expected results and what I have
Location quantity
LOCATION QUANTITY
Loc_1 20
Loc_2 10
SELECT BS.LOCATION, TL.PALL_NUM, BS.QUANTITY
FROM BATCH_STOCK BS
LEFT OUTER JOIN TRANS_LOG TL
ON TL.ITEM_NUM = BS.ITEM_NUM AND LOC_TO = BS.LOCATION
ORDER BY BS.LOCATION
Result
LOCATION PALL_NUM QUANTITY
Loc_2 16 5
Loc_2 15 5
Loc_2 14 5
Loc_1 13 5
Loc_1 12 5
Loc_1 11 5
Loc_1 10 5
... ...
Loc_1 1 5
This is just an extended version from the first question. But shows the output I would get if I didn't count backwards in the log. So for example Loc_2 would only have 2 pallets with quantity still on them but return 3 pallets because 3 pallets have been moved to that location. And Loc_1 would return 13 pallets but the first 9 have already been "picked" so they aren't there anymore.
Expected result
LOCATION PALL_NUM
Loc_2 16
Loc_2 15
Loc_1 13
Loc_1 12
Loc_1 11
Loc_1 10
Edit 2: Table structure
BATCH_STOCK
LOCATION = PK NVARCHAR(13) NOT NULL
QTY = NUMERIC (9,2) NOT NULL
TRANS_LOG
PALL_NUM = BIGINT NOT NULL
LOC_TO = NVARCHAR(16) NULL
QTY = NUMERIC(9,2) NULL

Answer 1

You're looking for a running total? This will return you the 4 rows:
select * from (
select *,
sum(QUANTITY) over (order by PALLET_NUMBER desc) as Quant
from yourtable
) X where Quant <= 20
That only works with full pallets, if you need to include rows for partial quantities, this should work then:
select * from (
select *, isnull(lag(Quant) over (order by PALLET_NUMBER desc),0) as NextQuant
from (
select *,
sum(QUANTITY) over (order by PALLET_NUMBER desc) as Quant
from yourtable
) X
) Y where NextQuant < 15
Update, added examples with the new tables included in the question
select * from (
select *,
sum(QUANTITY) over (partition by LOCATION order by PALL_NUM desc) as Quant
from #TRANS_LOG
) X join #BATCH_STOCK S on S.QUANTITY >= X.Quant and S.LOCATION = X.LOCATION
select * from (
select *, isnull(lag(Quant) over (partition by LOCATION order by PALL_NUM desc),0) as NextQuant
from (
select *,
sum(QUANTITY) over (partition by LOCATION order by PALL_NUM desc) as Quant
from #TRANS_LOG
) X
) Y join #BATCH_STOCK S on S.QUANTITY > Y.NextQuant and S.LOCATION = Y.LOCATION
Results:
   LOCATION PALL_NUM QUANTITY Quant LOCATION QUANTITY
1 Loc_1 13 5 5 Loc_1 20
2 Loc_1 12 5 10 Loc_1 20
3 Loc_1 11 5 15 Loc_1 20
4 Loc_1 10 5 20 Loc_1 20
5 Loc_2 16 5 5 Loc_2 10
6 Loc_2 15 5 10 Loc_2 10
   LOCATION PALL_NUM QUANTITY Quant NextQuant LOCATION QUANTITY
1 Loc_1 13 5 5 0 Loc_1 20
2 Loc_1 12 5 10 5 Loc_1 20
3 Loc_1 11 5 15 10 Loc_1 20
4 Loc_1 10 5 20 15 Loc_1 20
5 Loc_2 16 5 5 0 Loc_2 10
6 Loc_2 15 5 10 5 Loc_2 10
Example that I made

Answer 2

Okay, Here is what i have tried so far. I have used recursive CTE to achieve the expected output.
Schema
CREATE TABLE #tLocationQty
(
Location VARCHAR(10),
Quantity int
)
INSERT INTO #tLocationQty VALUES('Loc_1',20)
INSERT INTO #tLocationQty VALUES('Loc_2',10)
Second table schema
CREATE TABLE #tLog
(
MoveToLocation VARCHAR(10),
Quantity int,
PALL_NUM int
)
INSERT INTO #tLog VALUES('Loc_1',5, 13)
INSERT INTO #tLog VALUES('Loc_1',5, 12)
INSERT INTO #tLog VALUES('Loc_1',5, 11)
INSERT INTO #tLog VALUES('Loc_1',5, 10)
INSERT INTO #tLog VALUES('Loc_1',5, 9)
INSERT INTO #tLog VALUES('Loc_2',5, 16)
INSERT INTO #tLog VALUES('Loc_2',5, 15)
and below is the query.
Note: - I have not use order by clause. You have to use it as per your requirement.
Query
;WITH myCTE
AS
(
SELECT Location, Quantity AS Qty FROM #tLocationQty
UNION ALL
SELECT MoveToLocation, (Qty - t.Quantity) AS Qty FROM #tLog t
INNER JOIN myCTE
ON myCTE.Location = t.MoveToLocation
WHERE myCTE.Qty > 0
),
DistLocation
AS
(
SELECT DISTINCT Location FROM myCTE WHERE qty > 0
),
RowCnt
AS
(
SELECT Location, COUNT(DISTINCT Qty) AS Cnt FROM myCTE WHERE qty > 0
GROUP BY Location
)
--SELECT * FROM RowCnt
SELECT * FROM DistLocation
CROSS APPLY
(
SELECT TOP (SELECT top 1 Cnt FROM RowCnt r WHERE r.location = location)
PALL_NUM
FROM #tLog t
WHERE t.MoveToLocation = location
) c
Output
Location PALL_NUM
Loc_1 13
Loc_1 12
Loc_1 11
Loc_1 10
Loc_2 15
Loc_2 16