Translating a set theory idea into SQL [closed]

Question

It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form. For help clarifying this question so that it can be reopened, visit the help center.

Closed 9 years ago.

Say I have a the following db table:

╔═════════════════╦════════════╗
║ ADVERTISEMENTID ║ CATEGORYID ║
╠═════════════════╬════════════╣
║               1 ║ A          ║
║               1 ║ C          ║
║               2 ║ A          ║
║               2 ║ B          ║
║               3 ║ A          ║
╚═════════════════╩════════════╝

Given a list of categories passed as a parameter, say A, C I want to find only those advertisements that belong only to those two categories, in the above case only advertisement 1 would match.

Can anyone please help me translate this into SQL?

Answer 1

select advertismentid
from the_table
where categoryid in ('A', 'C')
group by  advertismentid
having count(*) = 2;

SQLFiddle: http://sqlfiddle.com/#!12/b94d6/1

This assumes that the same categoryid cannot be assigned more than once to the same advertismentid. It will also include advertisments that have A,C and other categories.

If you want those advertisments that have exactly categories A and C you need to exclude those that have more than that:

select advertismentid
from the_table
where categoryid in ('A', 'C')
group by  advertismentid
having count(*) = 2;
intersect
select advertismentid
from the_table
group by advertismentid
having count(*) = 2;

SQLFiddle: http://sqlfiddle.com/#!12/8901c/4
The SQLFiddle also has another solution using except instead of intersect

If your DBMS is limited and you cannot use except or intersect, you can use this alternative:

select t1.advertismentid
from the_table t1
where t1.categoryid in ('A', 'C')
group by t1.advertismentid
having count(*) = 2
and count(*) = (select count(*) 
                from the_table t2
                where t2.advertismentid = t1.advertismentid)

Answer 2

SELECT DISTINCT advertisementid 
FROM tft t1
WHERE t1.categoryid IN ('A','C')
AND EXISTS (
  SELECT * FROM tft t2
  WHERE t2.advertisementid = t1.advertisementid
  AND t2.categoryid IN ('A','C')
  AND t2.categoryid <> t1.categoryid
  );

Answer 3

It seems I came late, but here is my solution anyway:

SELECT advertisement
FROM   advertisement_childcare_types t1
LEFT JOIN (
       SELECT childcare_types ct
       FROM   table_childcare_types tct
       WHERE  childcare_types IN (0, 1, 3)
          ) AS mytypes
ON     t1.childcare_types = mytypes.ct
GROUP BY advertisement
HAVING SUM(IF(mytypes.ct IS NULL, -1, 1)) = 3;

You can test in your sqlfiddle using this modified version:

SELECT advertisement
FROM   advertisement_childcare_types t1
LEFT JOIN (SELECT 0 as ct UNION SELECT 1 UNION SELECT 3) AS mytypes
ON     t1.childcare_types = mytypes.ct
GROUP BY advertisement
HAVING SUM(IF(mytypes.ct IS NULL, -1, 1)) = 3;