Question

My current query:

    Select `userSubUnitsID`, `userSubUnits`.`userID`, CONCAT(`u`.`fname`,' ',`u`.`lname`) AS `full`, `pID`, SUM(`quantComp`) AS `total`
    From `userSubUnits`
    JOIN `u` ON `u`.`userID` = `userSubUnits`.`userID`
    GROUP BY `userID`

What I need is to get this by a date. So it will look more like this.

    +-----+------------+----------+-------------+
    |user |current week|prev week |two weeks ago|
    +-----+------------+----------+-------------+
    |John |         564|       354|          687|
    +-----+------------+----------+-------------+
    |Paul |         451|       328|          845|
    +-----+------------+----------+-------------+
    etc

Is this possible with a single query? I know how to limit it for the current week but not how to add in the previous weeks.

FYI, Weeks end Sunday at midnight.
Thanks in advance

Answer 1

SELECT user, SUM(WEEKOFYEAR(datefield) = WEEKOFYEAR(now()) AS current_week,
    SUM(WEEKOFYEAR(datefield) = WEEKOFYEAR(now() - INTERVAL 1 WEEK)) AS last_week,
    SUM(WEEKOFYEAR(datefield) = WEEKOFYEAR(now() - INTERVAL 2 WEEK)) AS two_weeks_ago
FROM ...
GROUP BY WEEKOFYEAR(now())

ugly, but should do the trick. Note that this will fail if your dates span a year boundary. To get around that, you’d have to add in some extra year-based logic.

As well, you’d be better off doing this kind of transformation client-side. It’s ugly to expand, and will very quickly get hideously inefficient if you need to start doing more than just a few weeks, e.g:

SELECT user, YEAR(datefield), WEEKOFYEAR(datefield), COUNT(*)
FROM ...
GROUP BY YEAR(datefield), WEEKOFYEAR(datefield)