Question

With a ‘standard’ Orders table I am trying to find out how far back in time it was that a specific client purchased a certain numbers of items.

Order ID      Items      Client      Date
--------      -----      ------      ----

1             1          Fred        26/04/2012
2             3          John        25/04/2012
3             2          Fred        20/04/2012
4             5          Fred        18/04/2012
5             3          Fred        14/04/2012
6             4          Fred        10/04/2012

So I want to know the time frame covered by the last 10 items purchased by Fred, starting from now and working back.

In this instance I would be trying to identify that Order IDs 1, 3, 4 & 5 added together take me to (or just past) my target total of 10 items, so the date I am looking for is 14/04/2012.

Is there a simple solution to this?

Answer 1

Off the top of my head…

CREATE FUNCTION time_of_last(itemcount INT, custid VARCHAR(50))
RETURNS DATE
BEGIN
   DECLARE tally INT DEFAULT 0;
   DECLARE ondate DATE;
   DECLARE curritems IN DEFAULT 0;
   DECLARE cur1 CURSOR FOR SELECT t.items, t.orderdate FROM yourtable t
      WHERE customer=custid ORDER BY orderdate DESC;
   DECLARE CONTINUE HANDLER FOR NOT FOUND SET ondate = NULL;

   OPEN cur1;

   read_loop: LOOP
       FETCH cur1 INTO curritems, ondate;
       SET tally = tally+curritems;
       IF (tally>=itemcount) THEN
            LEAVE read_loop;
       END IF;
   END LOOP;

   CLOSE cur1;

   RETURN ondate;
END;

SELECT time_of_last(10, 'Fred');

NB if you want to start gathering other information such as what orders are in the set, then Matt Fenwick’s solution is cleaner.