SQL Server : count how many times a record can be applied to a table

Question

I have two tables, Tickets and Discount . I have to count how many times the combination of a discount ID can be applied to orders in the Ticket table:

Tickets table:

ITEMID | CATEGORYID
-------+-----------
     1 |         11
     1 |         11
     1 |         11
     2 |         11
     3 |         12
     4 |         13
     4 |         13
     5 |         14

Discounts table:

DiscountID | ITEMID | CATEGORYID
-----------+--------+-----------
         1 |      1 |         11
         1 |      4 |         13
         2 |      1 |         11
         2 |      6 |         15
         3 |      1 |         11
         3 |      5 |         14

Result: the DiscountID 1 can be applied to the Ticket table twice, and DiscountID 3 can be applied once.

I currently manage to do this via loop in a stored procedure, it is slow as I have to loop through each discount items with all ticket items manually.

I wonder if there is a better way of doing it via SQL with combination of JOIN and INTERSECT . I couldn't get my head around this, I have to match more than 1 row in the discount table at the same time.

BETTER EXPLANATION:

Original Table without any discount: Tickets table:

ITEMID | CATEGORYID
-------+-----------
     1 |         11
     1 |         11
     1 |         11
     2 |         11
     3 |         12
     4 |         13
     4 |         13
     5 |         14

Apply Discount ID 1:

DiscountID | ITEMID | CATEGORYID
-----------+--------+-----------
         1 |      1 |         11
         1 |      4 |         13

Discount ID 1 counted twice because:

ITEMID | CATEGORYID
-------+-----------
     1 |         11
     1 |         11
     4 |         13
     4 |         13

Now onto Discount ID 2:

DiscountID | ITEMID | CATEGORYID
-----------+--------+-----------
         2 |      1 |         11
         2 |      6 |         15

With counted DiscountID 1 Items removed:

ITEMID | CATEGORYID
-------+-----------
     1 |         11
     2 |         11
     3 |         12
     5 |         14

Count for discount ID 2 is 0 because only one record of Dicount ID 2 can be matched in Ticket table 1 | 11

While

6 | 15 is not available in Ticket table. As Discount ID 2 isn't matched nothing is removed from Ticket table

Now onto Discount ID 3:

DiscountID | ITEMID | CATEGORYID 3 | 1 | 11

         3 |      5 |         14


ITEMID | CATEGORYID
-------+-----------
     1 |         11
     2 |         11
     3 |         12
     5 |         14

Discount ID 3 is counted once and now with counted records remove will be

ITEMID | CATEGORYID
-------+-----------
     2 |         11
     3 |         12

As I do not know how to execute above via SQL, I do not know how I can output:

DISCOUNTID | APPLIED
-----------+-----------
     1     |   2
     3     |   1

Answer 1

This will do the trick.

select distinct d.*
from discounts d
   inner join tickets t on t.itemId = d.itemid 
      and t.categoryId = d.categoryId

It will give you

DiscountID | ITEMID | CATEGORYID
-----------+--------+-----------
         1 |      1 |         11
         1 |      4 |         13
         2 |      1 |         11
         3 |      5 |         14

You can then sum up all the columns as necessary by

select discountid, count(*) as apply_discount_x_times from ( select distinct d.discountid, d.itemId, d.categoryId from discounts d inner join tickets t on t.itemId = d.itemid and t.categoryId = d.categoryId ) x group by discountid

Giving you

DiscountID | apply_discount_x_times -----------+-------- 1 | 2 2 | 1 3 | 2

Answer 2

select distinct d.*, CA.cnt
from discounts d
  CROSS APPLY 
    (SELECT count(0) cnt from tickets t2 
                                join discounts d2 
                                  on d.DiscountId = d2.discountid 
                                     and d2.itemid =  t2.itemid 
                                      and d2.categoryid = t2.categoryid) CA