How to manipulate VARRAYS in sql (oracle)? - sql

Question

Supposing i am using a table person, and persons might have multiple last names, so that attribute should be a varray of 3 elements for example (it's not about where to store last names), here is a simple sql for creating the type last name, the table person and adding an example row in oracle's sql developper (11G XE):
create type lastn as varray(3) of varchar2(10);
CREATE TABLE person
(
ID NUMBER NOT NULL
, last_name lastn
, CONSTRAINT EXEMPLE_PK PRIMARY KEY
(
ID
)
ENABLE
);
insert into person values(1,lastn('dani','bilel'));
I know how to update all last names at once, but i need to preserve existing last names and add other last names, or remove a single last name without affecting the others. In a nutshell, i want my code to be like (i am not familiar with PL/SQL):
insert into table
(select last_name from example where id=1)
values lastn('new');
This is the case where i want to get persons that have a first last name of 'bilel' and second last_name as 'dani'
select * from person where id in (select id from pernom p,table(p.last_name)
where column_value(1)='bilel' and column_value(2)='dani');
I know that it doesn't work like that, but i want to know CRUD(create update delete) statements in that case. and select statement with varray in where statement.
Thanks for your response.

Answer 1

From the docs:
Oracle does not support piecewise updates on VARRAY columns. However, VARRAY columns can be inserted into or updated as an atomic unit.
As shown in the examples there, you can manipulate the collection through PL/SQL instead; incuding adding an element to the array:
declare
l_last_name lastn;
begin
select last_name into l_last_name
from person where id = 1;
l_last_name.extend();
l_last_name(l_last_name.count) := 'third';
update person
set last_name = l_last_name
where id = 1;
end;
/
PL/SQL procedure successfully completed.
select last_name from person where id = 1;
LAST_NAME
--------------------------------------------------
LASTN('dani', 'bilel', 'third')
You can also do this via cast(multiset(...) as ...):
-- rollback; to reverse PL/SQL block actions above
update person p
set last_name = cast(multiset(
select column_value
from table (last_name)
union all
select 'third' from dual
) as lastn)
where id = 1;
1 row updated.
select last_name from person where id = 1;
LAST_NAME
--------------------------------------------------
LASTN('dani', 'bilel', 'third')
That explodes the existing last_name value into multiple rows, union's in a new value, and then converts the combined result back into your varray type.
And you can delete or update elements in a similar way:
update person p
set last_name = cast(multiset(
select column_value
from table (last_name)
where column_value != 'bilel'
) as lastn)
where id = 1;
1 row updated.
select last_name from person where id = 1;
LAST_NAME
--------------------------------------------------
LASTN('dani', 'third')
update person p
set last_name = cast(multiset(
select case column_value when 'third' then 'second' else column_value end
from table (last_name)
) as lastn)
where id = 1;
1 row updated.
select last_name from person where id = 1;
LAST_NAME
--------------------------------------------------
LASTN('dani', 'second')

Answer 2

For the select statement, i've figured out the solution, which goes like this :
select * from person p where id in (select id from table(p.last_name) where
column_value='bilel' intersect select id from table(p.last_name) where
column_value='dani');
or
select * from agent ag where id in (select id from table(ag.prenom)
t1,table(ag.prenom) t2,table(ag.prenom) t3 where t1.column_value='bilel' and
t2.column_value='dani' and t3.column_value='third');