Question

Consider the following table. It describes how many times user1 has started a conversation with user2 (“A started a conversation with B, 5 times”):

user1  user2  count
-------------------
A      B      5
A      C      2
B      A      6
B      C      1
C      A      9
C      B      4

However, I would like to merge rows between similar users. So user1:A and user2:B is the same as user1:B and user2:A, leading to the following result:

user1  user2  count
-------------------
A      B      11 (5 + 6)
A      C      11 (2 + 9)
B      C      5

My first thought was to SELECT the table to PHP, loop through it, add rows and INSERT the result back into a new table. But this seems very redundant (and slow, since the table contains thousands of records). Is this also possible to do with (My)SQL?

Answer 1

SELECT LEAST(user1, user2), GREATEST(user1, user2), SUM(count) AS count
FROM yourtable
GROUP BY LEAST(user1, user2), GREATEST(user1, user2)