Group rows by condition

Question

I have this data:

Start   End     Quantity
425     449     24
450     474     24
475     499     24
500     524     24
2300    2324    24
2400    2499    99
2500    2599    99
2800    2899    99
2900    2999    99
3200    3249    49
3250    3299    49
3300    3349    49
3350    3399    49
3400    3449    49
3500    3549    49
3600    3624    24
3650    3674    24
3700    3724    24
3950    3964    14
4000    4000    0
4150    4399    249
4400    4499    99
5034    5075    41

Quantity is a result of End - Start.

I would like to obtain the following data, the Generated rows:

Start   End     Quantity
425     449     24
450     474     24
475     499     24
500     524     24
425     524     96
2300    2324    24
2300    2324    24
2400    2499    99
2500    2599    99
-----GENERATED----
425     2599    438
------------------

2800    2899    99
2900    2999    99
3200    3249    49
3250    3299    49
3300    3349    49
3350    3399    49
3400    3449    49
3500    3549    49
-----GENERATED-----
2800    3549    492
------------------

3600    3624    24
3650    3674    24
3700    3724    24
3950    3964    14
4000    4000    0
4150    4399    249
4400    4499    99
5034    5075    41
-----GENERATED-----
3600    5075    475
------------------

The condition is that it has to sum all the quantities until 500. If it passes 500 do a new count.

I have tried with Rollup but I couldnt find the right condition to make it work.

Of course, this is way easier to do by programming code instead of SQL, but we must do it in database environment. The tools to get the generated rows can be anything, looping functions, new tables etc.

Error solving

I got into an error while running @Prdp's query:

Msg 530, Level 16, State 1, Line 1
The statement terminated. The maximum recursion 100 has been exhausted before statement completion.

I found the solution here: http://sqlhints.com/tag/the-statement-terminated-the-maximum-recursion-100-has-been-exhausted-before-statement-completion/

Update 1

Using @Prdp's query we got the following:

Start   End     rn  st
(400)   424     1   24
425     449     2   48
450     474     3   72
475     499     4   96
500     524     5   120
2300    2324    6   144
2400    2499    7   243
2500    2599    8   342
2800    (2899)  9   (441)  
(2900)  2999    10  99
3200    3249    11  148
3250    3299    12  197
3300    3349    13  246
3350    3399    14  295
3400    3449    15  344
3500    3549    16  393
3600    3624    17  417
3650    3674    18  441
3700    3724    19  465
3950    3964    20  479
4000    (4000)  21  (479) 
(4150)  4399    22  249
4400    4499    23  348
5034    (5075)  24  (389) 

Its getting closer to what we need. Would it be possible to extract only the data in between ( and ) while discarding the other data? We can use cursors too.

Answer 1

You can use Recursive CTE. I can't think of any better way.

;WITH cte
     AS (SELECT *,
                Row_number()OVER(ORDER BY start) rn
         FROM   Yourtable),
     rec_cte
     AS (SELECT *,
                ( [End] - Start ) AS st,
                1                 AS grp
         FROM   cte
         WHERE  rn = 1
         UNION ALL
         SELECT a.*,
                CASE
                  WHEN st + ( a.[End] - a.Start ) >= 500 THEN a.[End] - a.Start
                  ELSE st + ( a.[End] - a.Start )
                END,
                CASE
                  WHEN st + ( a.[End] - a.Start ) >= 500 THEN b.grp + 1
                  ELSE grp
                END
         FROM   cte a
                JOIN rec_cte b
                  ON a.rn = b.rn + 1)
SELECT Min(Start) as Start,
       Max([End]) as [End],
       Max(st) as Quantity
FROM   rec_cte
GROUP  BY grp
OPTION (maxrecursion 0)