Cursor and update in Postgresql - sql

Question

I've written a function in postgresql for updating a table using two cursors, the code executes fine without errors but doesn't do anything.
I have three tables
product master
sales data table
target table
So in the below function I'm getting product key from 1 and then in the sales table calculating no of stores/outlets the product was sold and then update the third table accordingly,
I need to calc the no of outlets for each product every week. but they should be unique, e.g If a product was sold in outlets A and B in first week the no of stores in first week should be 2, if it is sold in B, C, D next week the total should be 4 for 2nd week, If it is sold in A and D in third week the total should be still 4 for 3rd week. How do I do this.
Any help will be greatly appreciated.
create or replace function StCount() returns void as
$BODY$
Declare
ik1 integer ;
wk1 integer ;
Cur1 CURSOR FOR SELECT ik From table1 ;
Cur2 CURSOR FOR SELECT distinct(wk) FROM table2 order by wk asc;
begin
OPEN Cur1 ;
WHILE (Found) Loop
FETCH next from Cur1 INTO ik1 ;
OPEN Cur2 ;
WHILE (Found) Loop
FETCH next from Cur2 INTO wk1;
update table3 set skly =(select count(sk)from table2 a
where a.ik = ik1 and a.wk = wk1
and a.sk not in (select distinct (sk) from table2
where ik = ik1 and wk <= wk1 - 1 ))
where ik = ik1 and wk = w1
;
End loop ;
CLOSE Cur2;
end loop ;
CLOSE Cur1;
END;
$BODY$
LANGUAGE plpgsql VOLATILE;

Answer 1

You didn't show us how your tables look like or what exactly the data is you are dealing with, so I am assuming a lot here.
Using this setup:
create table product (id integer primary key);
create table sales (id serial primary key, product_id integer, store text, wk integer);
insert into product values (1),(2),(3),(4);
insert into sales (product_id, store, wk)
values
(1, 'A', 1),
(1, 'B', 1),
(1, 'B', 2),
(1, 'C', 2),
(1, 'D', 2),
(1, 'A', 3),
(1, 'D', 3);
The cumulative sales per product, store and week can be calculated like this:
select product_id,
store,
wk,
count(store) over (partition by product_id order by wk) as sofar
from (
select product_id,
store,
wk,
row_number() over (partition by product_id, store order by wk) as rn
from sales
order by product_id, store
) t
where rn = 1;
Note that the product table isn't actually used in this, but it can easily be joined to the inner query.
For the above sample data, this returns:
product_id | store | wk | sofar
-----------+-------+----+------
1 | A | 1 | 2
1 | B | 1 | 2
1 | C | 2 | 4
1 | D | 2 | 4
Now this result can be used to update the target table:
update target_table
set slky = t.sofar
from (
select product_id,
store,
wk,
count(store) over (partition by product_id order by wk) as sofar
from (
select product_id,
store,
wk,
row_number() over (partition by product_id, store order by wk) as rn
from sales
order by product_id, store
) t
where rn = 1
) s
where s.wk = target_table.wk
and s.product_id = target_table.product_id
and s.store = target_table.store;
This assumes that the combination of (product_id, store, wk) is unique in target_table. As you have obfuscated the names of your tables and columns in your example it is hard to tell though.