Nested Cursor

Question

Dears,
I have the below requirement:
Table A has columns a1 & a2, a3 and table B has column a1, b1 & b2. Retrieve a1, a2, a3 & sum of a3 for the same a1 values into resultset r1. Then iterate r1 & for each distinct a1, find the max value of b1 & the corresponding b2, put the result into resultset r2. Return r2
Please advise how I can achieve this using oracle procedure/function. This solution will help me to improve th performance of my application a lot

Answer 1

SQL> create table a (
  2    a1 integer,
  3    a2 varchar2(20),
  4    a3 number
  5  );
Table created.
SQL> insert into a(a1, a2, a3) values(1, '111', 123);
1 row created.
SQL> insert into a(a1, a2, a3) values(1, '22', 1);
1 row created.
SQL> insert into a(a1, a2, a3) values(1, '1', 2);
1 row created.
SQL> insert into a(a1, a2, a3) values(2, 'dd', 33);
1 row created.
SQL> insert into a(a1, a2, a3) values(3, 'ggg', 77);
1 row created.
SQL> insert into a(a1, a2, a3) values(3, 'jj', 999);
1 row created.
SQL> commit;
Commit complete.
SQL> select * from a;
        A1 A2                           A3
         1 111                         123
         1 22                            1
         1 1                             2
         2 dd                           33
         3 ggg                          77
         3 jj                          999
6 rows selected.
SQL> create table b(
  2    a1 integer,
  3    b1 number,
  4    b2 varchar2(20)
  5  );
Table created.
SQL> insert into b(a1, b1, b2) values(1, 22, 'ggg');
1 row created.
SQL> insert into b(a1, b1, b2) values(1, 2, 'g');
1 row created.
SQL> insert into b(a1, b1, b2) values(2, 5, '777');
1 row created.
SQL> insert into b(a1, b1, b2) values(2, 6, '555');
1 row created.
SQL> insert into b(a1, b1, b2) values(2, 7, '666');
1 row created.
SQL> insert into b(a1, b1, b2) values(3, 66, 'ggg');
1 row created.
SQL> insert into b(a1, b1, b2) values(3, 22, 'jjj');
1 row created.
SQL> commit;
Commit complete.
SQL> select * from b;
        A1         B1 B2
         1         22 ggg
         1          2 g
         2          5 777
         2          6 555
         2          7 666
         3         66 ggg
         3         22 jjj
7 rows selected.
SQL> with r1 as (
  2    select a.*,
  3      sum(a3) over(partition by a1) as sum_a3
  4    from a
  5  ),
  6  r2 as (
  7    select a1,
  8      max(b1) as b1,
  9      min(b2) keep (dense_rank last order by b1) as b2
10    from b
11    where a1 in (select a1 from r1)
12    group by a1
13  )
14  select * from r2;
        A1         B1 B2
         1         22 ggg
         2          7 666
         3         66 gggNot clear why do you need r1 if r2 is the only data to be returned. It can be calculated much simpler:
SQL> select a1,
  2    max(b1) as b1,
  3    min(b2) keep (dense_rank last order by b1) as b2
  4  from b
  5  where a1 in (select a1 from a)
  6  group by a1;
        A1         B1 B2
         1         22 ggg
         2          7 666
         3         66 gggRegards,
Dima
Message was edited by:
DimaCit