Sql Delete Statement Trouble - sql

Question

I am having trouble writing a script which can delete all the rows which match on the first three columns and where the Quantities sum to zero?
I think the query needs to find all Products that match and then within that group, all the Names which match and then within that subset, all the currencies which match and then, the ones which have quantities netting to zero.
In the below example, the rows which would be deleted would be rows 1&2,4&6.
Product, Name, Currency, Quantity
1) Product A, Name A, GBP, 10
2) Product A, Name A, GBP, -10
3) Product A, Name B, GBP, 10
4) Product A, Name B, USD, 10
5) Product A, Name B, EUR, 10
6) Product A, Name B, USD, -10
7) Product A, Name C, EUR, 10
Hope this makes sense and appreciate any help.

Answer 1

Try this:
DELETE
FROM [Product]
WHERE Id IN
(
SELECT Id
FROM
(
SELECT Id, SUM(Quantity) OVER(PARTITION BY a.Product, a.Name, a.Currency) AS Sm
FROM [Product] a
) a
WHERE Sm = 0
)

Answer 2

You may want to break this problem into parts.
First create a view that lists those combinations which sum to zero
CREATE VIEW vw_foo AS
SELECT product,name, currency, sum(quantity) as net
FROM foo
GROUP BY product, name, currency
HAVING sum(quantity)=0;
At this point, you need to make sure this view has the data you expect to delete. In you example, the view should have only 2 records: ProductA/NameA/GBP and ProductA/NameB/USD
Step 2. Delete the data where the fields match:
DELETE FROM foo
WHERE EXISTS
(SELECT *
FROM vw_foo
WHERE vw_foo.product = product
AND vw_foo.name = name
AND vw_currency = currency);

Answer 3

One way to simplify the SQL is to just concatente the 3 columns into one and apply some grouping:
delete from product
where product + name + currency in (
select product + name + currency
from product
group by product + name + currency
having sum(quantity) = 0)

Answer 4

I am assuming this is a accounting problem with offsetting pairs of entries in the ledger.
If there are for instance three entries for combination (A, A, GBP) this code and some of the example above will not work.
I create a temporary test table, loaded it with your data, used a CTE - common table expression - to find the duplicate pattern and joined it to the table to select the rows.
Just change the 'select *' to 'delete'.
Again, this only works for equal offsetting pairs. It will cause havoc with odd number of entries.
Do you have only even number of entries?
Sincerely
John
-- create sample table
create table #products
(
product_id int identity(1,1),
product_txt varchar(16),
name_txt varchar(16),
currency_cd varchar(16),
quantity_num int
);
go
-- add data 2 table
insert into #products
(product_txt, name_txt, currency_cd, quantity_num)
values
('A', 'A', 'GBP', 10),
('A', 'A', 'GBP', -10),
('A', 'B', 'GBP', 10),
('A', 'B', 'USD', 10),
('A', 'B', 'EUR', 10),
('A', 'B', 'USD', -10),
('A', 'C', 'EUR', 10);
go
-- show the data
select * from #products;
go
-- use cte to find combinations
with cte_Ledger_Offsets (product_txt, name_txt, currency_cd)
as
(
select product_txt, name_txt, currency_cd
from #products
group by product_txt, name_txt, currency_cd
having sum(quantity_num) = 0
)
select * from #products p inner join cte_Ledger_Offsets c
on p.product_txt = c.product_txt and
p.name_txt = c.name_txt and
p.currency_cd = c.currency_cd;