Selecting table to use in FROM clause based on certain condition - sql

Question

I have the following schema :
Purchase : pur_dt | pur_amt | item_code | quantity
ItemListForSoftware : item_code | item_desc | ...
ItemListForHardware : item_code | item_desc | ...
Now, I need to fetch item_desc from either of the Item tables based on the item_code of Purchase.
If item_code starts with S then it should look up in ItemListForSoftware
else from ItemListForHardware
Something like this
if(item_code starts with S){
select pur_dt,pur_amt,item_desc,quantity from Purchase, ItemListForSoftware where Purchase.item_code=ItemListForSoftware.item_code
}else{
select pur_dt,pur_amt,item_desc,quantity from Purchase, ItemListForHardware where Purchase.item_code=ItemListForHardware.item_code
}
Is there a way to do this via single SQL Query?

Answer 1

Is this what you want?
select p.pur_dt, p.pur_amt,
coalesce(ifs.item_desc, ifh.item_desc) as item_desc,
p.quantity
from Purchase p left join
ItemListForSoftware ifs
on p.item_code = ifs.item_code and
p.item_code like 'S%' left join
ItemListForHardware ifh
on p.item_code = ifh.item_code and
p.item_code not like 'S%';
Note the correct JOIN syntax.

Answer 2

You can solve this problem by LEFT JOINing to both tables and using the item_code to select the appropriate data:
SELECT p.pur_dt, p.pur_amt,
CASE WHEN LEFT(p.item_code, 1) = 'S' THEN s.item_desc
ELSE h.item_desc
END AS item_desc
p.quantity
FROM Purchase p
LEFT JOIN ItemListForSoftware s ON p.item_code=s.item_code
LEFT JOIN ItemListForHardware h ON p.item_code=h.item_code

Answer 3

Actually it is just as simple as you wrote:
select
pur_dt,
pur_amt,
case
when left(item_cide, 1) = 'S' then
(select item_desc from ItemListForSoftware as d where d.item_code = p.item_code)
else
(select item_desc from ItemListForHardware as d where d.item_code = p.item_code)
end as item_desc,
quantity
from Purchase as p;
Surely it will to work only if there is only 0 or 1 row in ItemListForSoftware and ItemListForHardware tables per item.
PS: Note that it could be faster a bit then joining three tables.