Insert multiple references in a nested table - sql

Question

I have the table customer_table containing a list (nested table) of references toward rows of the account_table.
Here are my declarations :
Customer type:
CREATE TYPE customer as object(
custid integer,
infos ref type_person,
accounts accounts_list
);
accounts_list type:
CREATE TYPE accounts_list AS table of ref account;
Table:
CREATE TABLE customer_table OF customer(
custid primary key,
constraint c_inf check(infos is not null),
constraint c_acc check(accounts is not null)
)
NESTED TABLE accounts STORE AS accounts_refs_nt_table;
So I would like to insert multiple refs in my nested table when I create a customer, as an account can be shared.
I can't find out how to do that.
I tried:
INSERT INTO customer_table(
SELECT 0,
ref(p),
accounts_list(
SELECT ref(a) FROM account_table a WHERE a.accid = 0
UNION ALL
SELECT ref(a) FROM account_table a WHERE a.accid = 1
)
FROM DUAL
FROM person_table p
WHERE p.personid = 0
);
With no success.
Thank you

Answer 1

You can use the collect() function, e.g. in a subquery:
INSERT INTO customer_table(
SELECT 0,
ref(p),
(
SELECT CAST(COLLECT(ref(a)) AS accounts_list)
FROM account_table a
WHERE accid IN (0, 1)
)
FROM person_table p
WHERE p.personid = 0
);
As the documentation says, "To get accurate results from this function you must use it within a CAST function", so I've explicitly cast it to your account_list type.
If you don't want a subquery you could instead do:
INSERT INTO customer_table(
SELECT 0,
ref(p),
CAST(COLLECT(a.r) AS accounts_list)
FROM person_table p
CROSS JOIN (SELECT ref(a) AS r FROM account_table a WHERE accid IN (0, 1)) a
WHERE p.personid = 0
GROUP BY ref(p)
);
but I think that's a bit messier; check the performance of both though...