unique output with select bulk and loop - oracle

Question

I have data structure looking like this:
So in column_a I have figures form 1 to 5, but in column_b I have elements which a repeating. And I want to loop column_b elements so that only unique are shown.
Here my pl/sql:
SET SERVEROUTPUT ON SIZE 100000
DECLARE
var_column_name_a VARCHAR2(30);
var_column_name_b VARCHAR2(30);
BEGIN
SELECT
column_name_a,
column_name_b
BULK COLLECT
INTO
var_column_name_a,
var_column_name_b
FROM
table_name;
FOR i IN var_column_name_a.first..var_column_name_a.last LOOP
dbms_output.put_line(var_column_name_b(i));
END LOOP;
END;
/
My current output is this:
type_a
type_b
type_b
type_b
type_c
but I need this
type_a
type_b
type_c
How to achieve this?

Answer 1

No way that code you posted produces anything, it is invalid. You can't bulk insert into scalar variables, so either you faked code you wrote, or the output.
Anyway: to me, the simplest option seems to be a separate select of distinct values into another collection. Here's how:
Sample data:
SQL> SELECT * FROM test;
COLA COLB
---------- ------
1 type a
2 type b
3 type b
4 type b
5 type c
Procedure:
SQL> DECLARE
2 TYPE tr IS RECORD
3 (
4 cola NUMBER,
5 colb VARCHAR2 (10)
6 );
7
8 TYPE tt IS TABLE OF tr;
9
10 l_tab tt; --> for all rows in TEST table
11 l_tab_unique SYS.odcivarchar2list; --> just for COLB
12 BEGIN
13 SELECT cola, colb
14 BULK COLLECT INTO l_tab
15 FROM test;
16
17 SELECT DISTINCT colb
18 BULK COLLECT INTO l_tab_unique
19 FROM test;
20
21 DBMS_OUTPUT.put_line ('all rows -------------');
22
23 FOR i IN l_tab.FIRST .. l_tab.LAST
24 LOOP
25 DBMS_OUTPUT.put_line (l_tab (i).cola || ' ' || l_tab (i).colb);
26 END LOOP;
27
28 DBMS_OUTPUT.put_line ('unique rows ----------');
29
30 FOR i IN l_tab_unique.FIRST .. l_tab_unique.LAST
31 LOOP
32 DBMS_OUTPUT.put_line (l_tab_unique (i));
33 END LOOP;
34 END;
35 /
The result is:
all rows -------------
1 type a
2 type b
3 type b
4 type b
5 type c
unique rows ----------
type b
type c
type a
PL/SQL procedure successfully completed.
SQL>

Answer 2

First off a simple select distinct col_b bulk ... is sufficient for what you are asking for. But assuming you may want a list of the other column you can get it with LISTAGG function. Also for this you can use an older (but still valid) form of BULK COLLECT where each column in the select is collected into a independent collection. So you wind up with the structure:
select col_a, col_b
bulk collect
into collection_a, collection_b
... ;
Specifically here with LISTAGG for col_a: (see demo)
declare
type array_t is table of varchar2(256) ;
k_tab constant varchar2(1) := chr(09);
col_a_list array_t;
col_b_list array_t;
begin
select col_b, listagg( col_a, ',') within group (order by col_a) a_list
bulk collect
into col_b_list, col_a_list --<<< BULK COLLECT into separate collection
from t1
group by col_b
order by col_b;
dbms_output.put_line ('col_b') ;
dbms_output.put_line ('_____');
for list_cnt in 1 .. col_b_list.count
loop
dbms_output.put_line(col_b_list(list_cnt));
end loop;
dbms_output.put_line (' ') ;
dbms_output.put_line ('col_b' || k_tab || 'col_a_list') ;
dbms_output.put_line ('_____' || k_tab || '__________') ;
for list_cnt in 1 .. col_b_list.count
loop
dbms_output.put_line(col_b_list(list_cnt) || k_tab || col_a_list(list_cnt));
end loop;
end ;