Eliminating duplicates in collection - oracle

Question

I have a declared collection of object type:
CREATE TYPE category_type AS OBJECT (
col1 VARCHAR2(6),
col2 VARCHAR2(10),
col3 NUMBER);
/
CREATE TYPE category_tab AS TABLE OF category_type;
/
And in the PL/SQL code:
v_category_data category_Tab := category_tab();
After I populate v_category_data, I find there are duplicates. I need to remove them. So, I have tried to work with MULTISET UNION DISTINCT by this:
v_new_data := v_category_data;
v_new_data := v_new_data MULTISET UNION DISTINCT v_category_data;
However, I get this error:
804/17 PLS-00306: wrong number or types of arguments in call to
'MULTISET_UNION_DISTINCT'
I tried using the MAP functionality, but did not understand how exactly to use it, or if it would help.
Does anyone know of a way to remove the duplicates in an object collection?

Answer 1

multiset union distinct requires the elements of the collection to be comparable. In your case the elements are PL/SQL records that are unfortunately not comparable data structures (i.e. PL/SQL provides no build-in mechanism to compare PL/SQL records).
multiset union works because it doesn't need to compare the elements.
If for some reason you do not want to use distinct when you're populating your v_category_data collection then I advise to do as I have written below. Selecting from the collection and using distinct on the columns.
This is the part that interests you:
SELECT Category_type(col1, col2, col3)
bulk collect INTO v_new_data
FROM (SELECT DISTINCT col1,
col2,
col3
FROM TABLE(v_category_data));
And this is the entire code with which I checked.
DECLARE
v_category_data CATEGORY_TAB := Category_tab();
v_new_data CATEGORY_TAB := Category_tab();
BEGIN
--populate the collection with 2 types of duplicates and print them
FOR i IN 1 .. 10 LOOP
v_category_data.Extend();
IF i < 4 THEN
V_category_data(v_category_data.last) :=
Category_type('Test', 'Test', 1);
ELSE
V_category_data(v_category_data.last) :=
Category_type('Test2', 'Test2', 2);
END IF;
END LOOP;
FOR i IN v_category_data.first..v_category_data.last LOOP
dbms_output.Put_line(V_category_data(i).col1
||' '
||V_category_data(i).col2
||' '
||V_category_data(i).col3);
END LOOP;
dbms_output.Put_line('After processing'
||Chr(10));
-- populate your collection using distinct and print the content
SELECT Category_type(col1, col2, col3)
bulk collect INTO v_new_data
FROM (SELECT DISTINCT col1,
col2,
col3
FROM TABLE(v_category_data));
FOR i IN v_new_data.first..v_new_data.last LOOP
dbms_output.Put_line(V_new_data(i).col1
||' '
||V_new_data(i).col2
||' '
||V_new_data(i).col3);
END LOOP;
END;

Answer 2

See below how we can achieve this :
CREATE TYPE category_type AS OBJECT (
col1 VARCHAR2(6),
col2 VARCHAR2(10),
col3 NUMBER);
/
CREATE OR REPLACE TYPE category_tab AS TABLE OF category_type ;
/
declare
v_category_data category_Tab := category_tab();
v_new_data category_Tab := category_tab();
begin
--Defining collection length
v_category_data.extend(3);
--Populating the collection
v_category_data(1):= category_type('A','B',1);
v_category_data(2):= category_type('A','B',1);
v_category_data(3):= category_type('B','C',2);
dbms_output.put_line('*********First Collection Elements**************');
for i in 1..v_category_data.count
loop
dbms_output.put_line(v_category_data(i).col1 || v_category_data(i).col2 || v_category_data(i).col3 );
end loop;
--Way to remove duplicate
select
v_category_data MULTISET UNION DISTINCT category_Tab(category_type('A','B',1))
into
v_new_data
from
dual;
--Displaying elements of second collection
dbms_output.put_line('*********Second Collection Elements**************');
for i in 1..v_new_data.count
loop
dbms_output.put_line(v_new_data(i).col1 || v_new_data(i).col2 || v_new_data(i).col3 );
end loop;
end;
Output
SQL> /
*********First Collection Elements**************
AB1
AB1
BC2
*********Second Collection Elements**************
AB1
BC2
PL/SQL procedure successfully completed.