Fusion of two tables with Specific SUM operation in SQL Server 2012 - sql

Question

I have two views in my database, here are their structures:
Table 1 (Stock product entries) :
---------------------------------------
| Date | Product | Quantity |
---------------------------------------
|2013-06-06| Procuct001 | 40 |
---------------------------------------
Table 2 (Stock product outputs) :
---------------------------------------
| Date | Product | Quantity |
---------------------------------------
|2013-06-07| Procuct001 | 15 |
---------------------------------------
|2013-06-08| Procuct001 | 5 |
---------------------------------------
I want to have a third view (or table) where I'll store all the stock's movements (entries or outputs) but I don't want to store the entered or retrieved quantity, but the difference (which means the stock's balance). In our case, the new table should contain :
Table 3 (Stock balance) :
---------------------------------------
| Date | Product | Quantity |
---------------------------------------
|2013-06-06| Procuct001 | 40 |
---------------------------------------
|2013-06-07| Procuct001 | 25 |
---------------------------------------
|2013-06-07| Procuct001 | 20 |
---------------------------------------
I am using SQL Server 2012 with SP1.

Answer 1

Here's one option using a Common Table Expression, utilizing ROW_NUMBER to seed your list (grouped by product):
with cte as (
select dt, product, quantity, row_number() over (partition by product order by dt) rn
from (
select * from t1
union
select * from t2
) t
)
select c.dt, c.product,
c2.quantity - coalesce(sum(c3.quantity),0) runningty
from cte c
inner join (
select product, quantity
from cte
where rn = 1) c2 on c.product = c2.product
left join cte c3 on c.product = c3.product and c3.rn <= c.rn and c3.rn <> 1
group by c.dt, c.product, c2.quantity
SQL Fiddle Demo
Depending on your data, you may not need to extra inner join to get your "1st" record -- you could perhaps use something like this instead:
with cte as (
select dt, product, quantity, row_number() over (partition by product order by dt) rn
from (
select * from t1
union
select * from t2
) t
)
select c.dt, c.product,
max(c.quantity) over (partition by c.product) - coalesce(sum(c3.quantity),0) runningty
from cte c
left join cte c3 on c.product = c3.product and c3.rn <= c.rn and c3.rn <> 1
group by c.dt, c.product, c.quantity