Advanced (My)SQL Query, is this possible? - php

Question

let me explain the whole thing with an example:
| id | product | rating |
1 23 54
2 23 54
3 23 53
4 24 33
5 26 22
6 24 11
Lets say we have multiple ratings for each product and want to display the three top products. This would mean we can can user Inner-/left-/right- Join to get the products name from another table, order it by desc and set a limit of 3. But this would show us the same product three times with a rating of 54, 54 and 53.
Is it possible to avoid products with the same id in the result just with SQL?
So the dream output from one SQL query would be:
| id | product | rating |
1 23 54
4 24 33
5 26 22
In words: the top three unique products by rating (and of course only the row of the item with the highest rating -> id 1 or 2 for product 23 and not id 3).
Further more if there is only one product or two products with multiple ratings it should only transfer 1 or 2 results.

Answer 1

You can do this by taking the maximum rating for each product and choosing the top three:
select product, max(rating) as maxrating
from table t
group by product
order by maxrating desc
limit 3;
If you want the id for this rating, you can use the substring_index()/group_concat() trick:
select product, max(rating) as maxrating,
substring_index(group_concat(id order by rating desc), ',', 1) as id
from table t
group by product
order by maxrating desc
limit 3;
Alternatively, you can eschew the group by:
select t.*
from table t
where not exists (select 1
from table t2
where t2.product = t.product and
(t2.rating > t.rating or
t2.rating = t.rating and t2.id > t.id
)
)
order by t.rating desc
limit 3;
The complicated where clause is because multiple ratings can be the same.
EDIT:
The not exists version is getting the highest rating on the highest id for each row. The logic is simply saying: "Get me all rows from the products table where the product in the row has no other row with a higher rating/id combination". This is an awkward way for people to understand "Get the row with the maximum rating". But it turns out to be easier for the database to process. It is typically the most efficient method in MySQL and often the most efficient method in other databases as well, particularly with the right indexes defined.

Answer 2

Use a SELECT DISTINCT query. Check out details here: http://dev.mysql.com/doc/refman/5.6/en/select.html