How to select multiple columns without duplicates in one column in Oracle

The problem: I need find top client for supplier. Top client has max items sum on all orders. If supplier has 2 top clients, then top client will be with min order id, which means - who first created order and get max items - it will be top. I was writing this sql code:

select s.s_id, c.c_id, min(o.o_id), count(*)
from suppliers s, clients c, orders o, items i
where s.s_id=c.id_s and c.c_id=o.id_c and o.o_id=i.id_o
group by s.s_id, c.c_id
order by 4 desc, 3

and get this result: http://imageshack.us/photo/my-images/148/32969388.jpg/

but i need to 开发者_如何学编程get: http://imageshack.us/photo/my-images/842/51815927.jpg/

---

You need to rank the order count and ID. So you should use an analytic and inline views. Something like:

select s_id
       , c_id
       , min_order_id
       , no_of_orders
from (
    select s_id
           , c_id
           , min_order_id
           , no_of_orders
           , rank() over (partition by s_id 
                          order by no_of_orders DESC, min_order_id ASC) rnk
    from (
        select s.s_id
               , c.c_id
               , min(o.o_id) as min_order_id
               , count(*) as no_of_orders
        from suppliers s, clients c, orders o, items i
        where s.s_id=c.id_s and c.c_id=o.id_c and o.o_id=i.id_o
        group by s.s_id, c.c_id
        )
    )    
where rnk=1
/