Need help in framing an SQL query

Question

Hi ,
Requirement is that
Let’s say I have a table say X
I am passing two input params from UI and in DAO i am querying these input prams against the table X(has c1,c2 as columns ) (as it has two columns mapping the input params A,B from UI).
if either one or Both of the input params matches ,only then i should fetch the records accordingly. It's understood that there won't be a scenario that both input params are null from UI.
I have three conditions here,
1.     A match against X, B doesnt Match
2.     B match against X, A doesnt Match
3.     Both A & B match against X
if i use let's say,
SELECT *FROM X WHERE (c1= A OR c2= B) .
•     if A match against X B doesnt Match It is fetching all the records matching A - requirement fullfilled
•     B match against X A doesnt Match It is fetching all the records matching B- requirement fullfilled.
•     Both A & B match against X It is fetching all the records from the table matching the first condition (which here is A) . requirement not fullfilled.
I am thinking of using the following query but I need to frame it in a way that based on ACHK & BCHK we need to add the conditions.
SELECT CASE WHEN C1 = A THEN 0 ELSE 9 END AS ACHK, CASE WHEN C2 = B THEN 0 ELSE 9 END AS BCHK FROM X
Based on ACHK & BCHK append the conditions. i.e.
IF ACHK = 0 AND BCHK = 0 THEN
AND C1 = A AND C2 = B
ELSE
AND (C1 = A OR C2 = B)
Please help.

Answer 1

Hi,
963022 wrote:
Would be glad if you can shed some light on role of DENSE_RANK here (ofcourse i did read the forums but i am confused on what basis a row is ranked).Complicated queries and expressions are usually built of smaller, simpler parts. To understand how the big, complicated thing works, look first at the smaller, simpler components of it, and make sure you understand how they work.
In this case, the query is composed of a sub-query and a main query. Start by looking at the sub-query only. Display all rows and all columns of the sub-query.
Break apart the complicated components of that sub-query. Rnk depends on an ORDER BY clause, but the expression used in that ORDER BY clause isn't displayed anywhere. Copy and display just that expression. (I called this case_1_2 in the query below.) Even that may be too complicated to understand, so break it apart. Case_1_2 is formed by adding 2 smaller CASE expressions; in the query below, I displayed each of the in separate columns (case_1 and case_2).
So here's a query you might use to understand and debug the query I posted:
WITH     targets          AS
     SELECT     'ABC' AS c1, 10 AS c2  FROM dual     UNION ALL
     SELECT     'ABC',          20            FROM dual     UNION ALL
     SELECT     'FOO',          10            FROM dual     UNION ALL
     SELECT     'FOO',          20            FROM dual     UNION ALL
     SELECT     'FOO',          99            FROM dual
,     got_rnk          AS
     SELECT     x.c1, x.c2
     ,     t.c1          AS target_c1
     ,     t.c2          AS target_c2
     ,     DENSE_RANK () OVER ( PARTITION BY  t.c1, t.c2
                               ORDER BY          CASE
                                             WHEN  x.c1  = t.c1
                                     THEN  1
                                     ELSE  0
                                         END
                               + CASE
                                             WHEN  x.c2  = t.c2
                                     THEN  2
                                     ELSE  0
                                         END
                         DESC
                       )          AS rnk
--     *****  DEBUGGING SECTION  *****     The following columns are included only
--                                       to see how the function above works.
     ,                               CASE
                                             WHEN  x.c1  = t.c1
                                     THEN  1
                                     ELSE  0
                                         END
                               + CASE
                                             WHEN  x.c2  = t.c2
                                     THEN  2
                                     ELSE  0
                                         END           AS case_1_2
     ,                               CASE
                                             WHEN  x.c1  = t.c1
                                     THEN  1
                                     ELSE  0
                                         END           AS case_1
     ,                            CASE
                                             WHEN  x.c2  = t.c2
                                     THEN  2
                                     ELSE  0
                                         END           AS case_2
--     *****  END OF DEBUGGING SECTION  *****
     FROM     targets  t
     JOIN           x  ON   x.c1     = t.c1
                   OR   x.c2     = t.c2
--     The following main query is only for debugging
SELECT       *
FROM       got_rnk
ORDER BY  target_c1, target_c2
,            rnk
;Output:
              TARGET TARGET          CASE
C1         C2 _C1       _C2    RNK   _1_2 CASE_1 CASE_2
ABC        10 ABC        10      1      3      1      2
ZZZ        10 ABC        10      2      2      0      2
ABC         5 ABC        10      3      1      1      0
ABC         6 ABC        10      3      1      1      0
ABC         7 ABC        10      3      1      1      0
ABC         1 ABC        10      3      1      1      0
ABC         3 ABC        10      3      1      1      0
ABC         2 ABC        10      3      1      1      0
ABC         8 ABC        10      3      1      1      0
ABC         9 ABC        10      3      1      1      0
ABC         4 ABC        10      3      1      1      0
ZZZ        20 ABC        20      1      2      0      2
ABC        10 ABC        20      2      1      1      0
ABC         2 ABC        20      2      1      1      0
ABC         1 ABC        20      2      1      1      0
ABC         9 ABC        20      2      1      1      0
ABC         8 ABC        20      2      1      1      0
ABC         7 ABC        20      2      1      1      0
ABC         6 ABC        20      2      1      1      0
ABC         5 ABC        20      2      1      1      0
ABC         4 ABC        20      2      1      1      0
ABC         3 ABC        20      2      1      1      0
ABC        10 FOO        10      1      2      0      2
ZZZ        10 FOO        10      1      2      0      2
ZZZ        20 FOO        20      1      2      0      2Study this output to see how the different components work.
For example, the first 12 rows of output show what we found when we were looking for ('ABC', 10). There were 12 rows in table x that had either ca='ABC' or c2=10 (or both). We used the rnk column to find the best match: lower numbers mean better matches. How did we do that? By using the DENSE_RANK function with a DESCending order by clause, that is, we used <tt> DENSE_RANK () OVER (... ORDER BY exp DESC) </tt>, which means that higher values of exp produce lower values from DENSE_RANK. What is exp? It's the column I called case_1_2. As you can see, when there was a perfect match ('ABC', 10), then case_1_2 had the value 3. Less perfect matches got lower values of case_1_2; for example ('ZZZ', 10) only had 2 in the case_1_2 column, and ('ABC', 5) only had 1. Why did case_1_2 have values of 3, 2 or 1 on those rows? Well, it's the sum of case_1 and case_2, so look at those individual columns, and see how they are computed.