Imagine you've got a table with products. Each product has a price and belongs to a specific category. Furthermore, each product also belongs to a sub-category. Now, what if you want to find the cheapest product for each sub-category? That's easy:
开发者_运维问答SELECT MIN(price), sub_category FROM products GROUP BY sub_category
Right?
Now, imagine you only want to show a maximum of 2 products for each category (not sub-category) for the above result. Is there a way to achieve that in the SQL?
Assuming the products have a unique id you may try this:
select * from products p
where p.id in (select p1.id from products p1
where p.category = p1.category order by price limit 0,2)
and price = (select min(price) from products p2
where p2.sub_category = p.sub_category)
EDIT: @wimvds: Thanks for pointing out the mistake.
EDIT 2: I guess this is still wrong if there is more than one product with a minimal price in a subcategory.
Shouldn't it be like this:
select * from products p, products p1 where p.id in (select min(price) from products p2 where p2.sub_category = p2.sub_category) and p1.category=p.category order by p.price limit 0,2
精彩评论