Question

I have a table called Sales:

id_sale  id_ticket  total  state
-------  ---------  -----  -----
100      100        30     inactive    
101      101        30     active    
102      101        60     active    
103      102        30     active    
104      102        30     active    
105      103        10     active    
106      103        5      active    

What I need is to get the total sum of the total column, sum(total) when the state is active and multiply it for x percentage (I can do that), having that I have to select all the tickets that their sum (of all the tickets summed) is <= sum(total) * x percentage and copy it to another table (same values)

So if I have the sum of ticket 101 = 90 and the sum of ticket 102 = 60 and the sum of ticket 103 = 15, and the sum(total)*x percentage = 160, the query will only return all the info from the tickets 101 and 102.

Reading all morning I found a code that somehow does something like what I need but using id_Sales and not id_ticket:

    select t1.id_sale, t1.total, SUM(t2.total) as sum
    from sales t1
    inner join sales t2 on t1.id_sale >= t2.id_sale
    where t2.state='active'
    group by t1.id_sale, t1.total
    having sum(t2.total)<=(
                           select sum(total)*.65
                           from sales
                           where state='active')
    order by t1.id_sale

What i espect to recieve is:

id_sale  id_ticket  total  state
-------  ---------  -----  ----- 
101      101        30     active    
102      101        60     active    
103      102        30     active    
104      102        30     active

In wich the sum of ticket 101 and ticket 102 is <= sum(total)*x

so the sum of 101 and 102 = 150 and sum(total)*x = 160 (for example)

How can I do what I need or how can I edit this code so it does what I’m looking for.

Answer 1

You could start with something like this:

WITH grouped_and_ranked AS (
  SELECT
    id_ticket,
    ticket_total = SUM(total),
    grand_total = SUM(SUM(total)) OVER (),
    ticket_rnk = ROW_NUMBER() OVER (ORDER BY id_ticket)
  FROM sales
  WHERE state = 'active'
  GROUP BY id_ticket
),
cumulative AS (
  SELECT
    id_ticket,
    ticket_total,
    grand_total,
    ticket_rnk,
    total_so_far = ticket_total
  FROM grouped_and_ranked
  WHERE ticket_rnk = 1
  UNION ALL
  SELECT
    r.id_ticket,
    r.ticket_total,
    r.grand_total,
    r.ticket_rnk,
    total_so_far = c.total_so_far + r.ticket_total 
  FROM grouped_and_ranked r
    INNER JOIN cumulative c ON r.ticket_rnk = c.ticket_rnk + 1
  WHERE r.ticket_total columns
)
SELECT
  s.id_sale,
  s.id_ticket,
  s.total,
  s.state
FROM sales s
INNER JOIN cumulative c ON s.id_ticket = c.id_ticket
;

The first CTE, grouped_and_ranked, retrieves active sales and calculates totals by ticket as well as the grand total (using windowed SUM()). It also assigns ranking numbers to the tickets to be used later when calculating the running total.

The next CTE, cumulative, is a recursive CTE. It retrieves rows from the previous result set as long as the running total is no greater than the specified @percent‘age of the grand total.

The main SELECT uses the final ticket list to get the detail rows, by joining the list back to the sales table.

This query can be tested at SQL Fiddle.