Question

I am wishing to compare two sets of data using SQL wherever possible. Please consider the following data structures:

tbl_users:
+---------+----------+
| user_id |   avatar |
+---------+----------+
|    1    | test.jpg |
|    2    | 1234.jpg |
+---------+----------+

tbl_shortlists:
+------+--------+
|  id  |  owner |
+------+--------+
|   1  |    1   |
|   2  |    2   |
+------+--------+

tbl_shortlist_items:
+---------+--------------+
| item_id | shortlist_id |
+---------+--------------+
|    1    |       1      |
|    2    |       1      |
|    3    |       1      |
|    1    |       2      |
|    2    |       2      |
+---------+--------------+   

I wish to select the tbl_users.user_id and tbl_users.avatar where the shortlist owned by any user contains two or more of the same item_id as the current owner. So, let’s assume I am looking up the user_id 1, with the above data structure. Since the Shortlist with the ID shares two items with shortlist_id 1 (i.e. user 1’s shortlist), I would like to return user_id = 2 and avatar = 1234.jpg.

I am at a loss as to how to do this in pure SQL. I was thinking that it might be possible to use IN(), but I don’t know if that would work.

Here’s some pseudo-code to hopefully explain a little better what I would like:

Select user_id and avatar
for each shortlist that contains two or more item_ids that are in the shortlist
that is owned by user_id = 1.

Answer 1

As always with SQL it helps to break it down into small sections. i.e. the first thing you need is all the items in short lists that the current user owns:

SELECT  Item_ID
FROM    tbl_ShortList_Items
        INNER JOIN tbl_ShortLists
            ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
WHERE   Owner = 1 -- CURRENT USER

Next you need all users who have short lists with the same items in. For explanatory purposes I will build up as an IN statement, but an INNER JOIN may perform better.

SELECT  Owner
FROM    tbl_ShortList_Items
        INNER JOIN tbl_ShortLists
            ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
WHERE   Item_ID IN 
        -- RESULTS FROM LAST QUERY START
        (   SELECT  Item_ID
            FROM    tbl_ShortList_Items
                    INNER JOIN tbl_ShortLists
                        ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
            WHERE   Owner = 1 -- CURRENT USER
        )   
        -- RESULTS FROM LAST QUERY END
AND     Owner != 1 -- CURRENT USER

You then need to limit this to those that have 2 or more, by adding a GROUP BY and HAVING

SELECT  Owner
FROM    tbl_ShortList_Items
        INNER JOIN tbl_ShortLists
            ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
WHERE   Item_ID IN 
        -- RESULTS FROM LAST QUERY START
        (   SELECT  Item_ID
            FROM    tbl_ShortList_Items
                    INNER JOIN tbl_ShortLists
                        ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
            WHERE   Owner = 1 -- CURRENT USER
        )   
        -- RESULTS FROM LAST QUERY END
AND     Owner != 1 -- CURRENT USER
GROUP BY Owner
HAVING  COUNT(DISTINCT tbl_ShortList_Items.Item_ID) > 1

You then need to use tbl_users to get the avatar for these owners

SELECT  User_ID, Avatar
FROM    tbl_Users
WHERE   User_ID IN 
        -- RESULTS FROM LAST QUERY START
        (   SELECT  Owner
            FROM    tbl_ShortList_Items
                    INNER JOIN tbl_ShortLists
                        ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
            WHERE   Item_ID IN 
                    (   SELECT  Item_ID
                        FROM    tbl_ShortList_Items
                                INNER JOIN tbl_ShortLists
                                    ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
                        WHERE   Owner = 1 -- CURRENT USER
                    )
            AND     Owner != 1 -- CURRENT USER
            GROUP BY Owner
            HAVING  COUNT(DISTINCT tbl_ShortList_Items.Item_ID) > 1
        )
        -- RESULTS FROM LAST QUERY END

As I said, I think rearranging this to JOINs will be optimised better, but have not tested this theory.

SELECT  User_ID, Avatar
FROM    tbl_Users
        INNER JOIN
        (   SELECT  Owner
            FROM    tbl_ShortList_Items
                    INNER JOIN tbl_ShortLists
                        ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
                    INNER JOIN
                    (   SELECT  Item_ID
                        FROM    tbl_ShortList_Items
                                INNER JOIN tbl_ShortLists
                                    ON tbl_ShortLists.ID = tbl_ShortList_Items.ShortList_ID
                        WHERE   Owner = 1 -- CURRENT USER
                    ) CurrentUserItems
                        ON CurrentUserItems.Item_ID = tbl_ShortList_Items.Item_ID
            WHERE   Owner != 1
            GROUP BY Owner
            HAVING  COUNT(DISTINCT tbl_ShortList_Items.Item_ID) > 1
        ) MatchUsers
            ON MatchUsers.Owner = tbl_Users.User_ID