Optimize sql query with the rank function

Question

This query gets the top item in each group using the ranking function.

I want to reduce the number of inner selects down to two instead of three. I tried using the rank() function in the innermost query, but couldn't get it working along with an aggregate function. Then I couldn't use a where clause on 'itemrank' without wrapping it in yet another select statement.

Any ideas?

select *
from (
    select 
        tmp.*,
        rank() over (partition by tmp.slot order by slot, itemcount desc) as itemrank
    from (
        select
            i.name, 
            i.icon,
            ci.slot,
            count(i.itemid) as itemcount
        from items i
        inner join citems ci on ci.itemid = i.itemid
        group by i.name, i.icon, ci.slot    
    ) as tmp
) as popularitems
where itemrank = 1

EDIT: using sql server 2008

Answer 1

In Oracle and Teradata (and perhaps others too), you can use QUALIFY itemrank = 1 to get rid of the outer select. This is not part of the ANSI standard.

Answer 2

You can use Common Table Expressions in Oracle or in SQL Server.

Here is the syntax:

WITH expression_name [ ( column_name [,...n] ) ]
AS
( CTE_query_definition )

The list of column names is optional only if distinct names for all resulting columns are supplied in the query definition.

The statement to run the CTE is:

SELECT <column_list>
FROM expression_name;