Question

Consider the following table:

+---------+---------+------------+----------+------------+
| msg_id  | user_id | _date      | _time    | count_for  |
+-------------------+------------+----------+------------+
|   1     | 1       | 2011-01-22 | 06:23:11 | 2011-01-22 |
|   2     | 1       | 2011-01-23 | 16:17:03 | 2011-01-24 |
|   3     | 1       | 2011-01-23 | 17:05:45 | 2011-01-24 |
|   4     | 2       | 2011-01-24 | 06:58:13 | 2011-01-24 |
|   5     | 2       | 2011-01-24 | 15:59:32 | 2011-01-24 |
|   6     | 3       | 2011-01-25 | 13:45:00 | ...        |
|   7     | 3       | 2011-01-26 | 13:22:34 | ...        |
+---------+---------+------------+----------+------------+

I want to count the number of messages per day, but messages after 16:00 this day have to count for the next day:
– msg 1 is written before 16:00 on 2011-01-22, so count_for is (same day) 2011-01-22
– msg 2 is written after 16:00 on 2011-01-23, so count_for is (next day) 2011-01-24

Obviously, I did the first 5 manually. My question is: with what query can I perform this operation automatically (in MySQL)?

Answer 1

Something like

SELECT count(msg_id), if(_time >= '16:00:00', _date + INTERVAL 1 DAY, _date) AS effective_date
FROM yourtable
GROUP BY effective_date

should do the trick.