Sql query-filter if 1 set of data exists and in the same time 2 set doesnt

Question

Hi guys,
I need to write a query where I filter on certain data being present, but in the same time other data from the same table being not present.
For example:
ProceduresTable:
Type 1
Type 1
Type 2
Type 2
Type 1
MyTable:
Name
OtherFields
JoiningTable (many to many)
ProceduresTable_Key
MyTable_Key
I need to select all records from MyTable and Procedures for which there are procedures from Type 1 but there are no procedures from Type 2 in the same time.
So far I've been only able to do that by using subqueries and I think there should be an easier way to do that.
Here is what I have:
Select * From MyTable, Procedures, JoiningTable Where JOINING_CONDITIONS AND Type = 1 AND NAME_SOME_COLUMN NOT IN (SELECT NAME_SOME_COLUMN FROM ... WHERE Type = 2 and inner.Name = outer.Name)
Hope you understand my meta code. I have Oracle 9 db.
Thanks a lot!

Answer 1

Hi,
Welcome to the forum!
Whenever you have a question, it helps if you post a specific example. Include CREATE TABLE and INSERT statements for a little sample data, and also post the results you want from that data.
if you can use commonly available tables (like those in the scott or hr schemas) to illustrate your proble, then you don't have to post the sample data; just the results.
For example, I think you're asking something like this:
"How can I find information about people in the sciott.emp table who are in a department that has at least one CLERK ( that is, a row with job='CLERK') but no SALESMAN? That is, I want these results:
{code}
ENAME DEPTNO JOB
CLARK 10 MANAGER
KING 10 PRESIDENT
MILLER 10 CLERK
JONES 20 MANAGER
FORD 20 ANALYST
ADAMS 20 CLERK
SMITH 20 CLERK
SCOTT 20 ANALYST
{code}
One way is to use CASE expressions instead of WHERE clauses to test for the criteria, then using the reults of the CASE expressions in the WHERE clause.
For example:
WITH     got_cnts     AS
     SELECT     ename,     deptno,     job
     ,     COUNT ( CASE
                         WHEN  job = 'CLERK'
                   THEN  1
                     END
                ) OVER (PARTITION BY deptno)     AS clerk_cnt
     ,     COUNT ( CASE
                         WHEN  job = 'SALESMAN'
                   THEN  1
                     END
                )     OVER (PARTITION BY deptno)     AS salesman_cnt
     FROM     scott.emp
SELECT     ename,     deptno,     job
FROM     got_cnts
WHERE     clerk_cnt     > 0
AND     salesman_cnt     = 0
;The problem here is that you can't always tell, by looking at any one row, if it should be included or not; in this case, you need to know something about the department as a whole. Analytic functions can look at the department as a whole, but analytic functions are computed after the WHERE clause is applied, so to use the results of the analytic functions in a WHERE clause, we need to do the analytics first, in a sub-query, and the WHERE clause later.
Sometimes (as in the example above) analytic functions are useful; other times, aggregate functions are more appropriate, depending on the exact requirements.
Edited by: Frank Kulash on Apr 12, 2010 9:00 AM