选择 MYSQL 中具有最大值的所有行

Question

table for visualization

item     title    numTimesPurchased
1        Beer     30
2        Chips    13
3        Smokes   30
4        Gum      3

Code I've written so far

SELECT Purchase.item, Item.title, SUM(quantity) AS numTimesPurchased 
FROM Item 
INNER JOIN Purchase ON Item.id = Purchase.item 
GROUP BY item, title;

I need to select all columns in the row/ multiple rows where which has a MAX value in numTimePurchased. So in this table, both item 1 and 3 should be selected.

I've tried sorting the data using ORDER BY and LIMIT but it will not work with data sets where there are more than one row with a max value, is there another way to approach this problem?

Answer 1

One nice way to handle this on MySQL 8+ uses the RANK analytic function:

WITH cte AS (
    SELECT p.item, i.title, SUM(quantity) AS numTimesPurchased,
           RANK() OVER (ORDER BY SUM(quantity) DESC) rnk
    FROM Item i
    INNER JOIN Purchase p ON i.id = p.item
    GROUP BY p.item, i.title
)

SELECT item, title, numTimesPurchased
FROM cte
WHERE rnk = 1;

A way to handle this on earlier versions of MySQL uses a non correlated subquery in the HAVING clause to check the number of times purchased:

SELECT p.item, i.title, SUM(quantity) AS numTimesPurchased
FROM Item i
INNER JOIN Purchase p ON i.id = p.item
GROUP BY p.item, i.title
HAVING SUM(quantity) = (SELECT SUM(quantity)
                        FROM Item i
                        INNER JOIN Purchase p ON i.id = p.item
                        GROUP BY p.item, i.title
                        ORDER BY SUM(quantity) DESC
                        LIMIT 1);

Answer 2

WITH cte AS ( SELECT Purchase.item, Item.title, SUM(quantity) AS numTimesPurchased 
                   , MAX(SUM(quantity)) OVER () maxsum
              FROM Item 
              INNER JOIN Purchase ON Item.id = Purchase.item 
              GROUP BY item, title )
SELECT * FROM cte WHERE numTimesPurchased = maxsum

MAX(SUM(quantity)) OVER () in CTE calculates maximal value (of SUM(quantity) = numTimesPurchased ) within all rows. Outer query hence filters the rows where numTimesPurchased is not equal to this maximal value (formally not less needed, but the value above maximal value cannot exist).

PS. I hope that quantity is not of float or double datatype. But if it is then SUM(quantity) (both expressions in CTE) must be rounded to reasonable accuracy.