Populate an Associative Array inside a Nested Table - sql

Question

In relation to my Previous Question, Include RowId value in Nested Table,
I have the below sample table:
create table data_test
(
data_id number,
data_value varchar2(100),
batch_name varchar2(100)
);
I have used this table as a parameter which includes the rowid:
create or replace package dat_pkg is
type typ_dat_rec is record (
data_rec data_test%rowtype,
data_rowid rowid);
type typ_dat_tst is table of typ_dat_rec index by pls_integer;
procedure transform_dat (p_batch_name data_test.batch_name%type);
procedure proc_test (p_dat typ_dat_tst);
end dat_pkg;
/
Using the procedure transform_dat, I want to populate the variable l_dat_rec with the filtered records from the data_test table, transform the data, and finally update the records using the proc_test procedure:
create or replace package body dat_pkg is
procedure transform_dat (p_batch_name data_test.batch_name%type)
is
cursor cur_dat is
select rowid, a.*
from data_test a
where batch_name = p_batch_name;
l_dat_rec typ_dat_tst;
begin
open cur_dat;
fetch cur_dat
BULK COLLECT
into l_dat_rec;
close cur_dat;
-- Do the Transformation here. Example --
for i in l_dat_rec.count loop
if l_dat_rec(i).data_value = 'hello' then
l_dat_rec(i).data_id := l_dat_rec(i).data_id + l_dat_rec(i).data_id;
else
l_dat_rec(i).data_id := l_dat_rec(i).data_id * l_dat_rec(i).data_id;
end if;
end loop;
-- update the table
proc_test (p_dat => l_dat_rec);
end transform_dat;
procedure proc_test (p_dat typ_dat_tst)
is
begin
for i in 1..p_dat.count loop
update data_test
set data_value = p_dat(i).data_value
where data_id = p_dat(i).data_id
and rowid = p_dat(i).data_rowid;
end loop;
end proc_test;
end dat_pkg;
/
however i am getting the error PLS-00597: expression 'L_DAT_REC' in the INTO list is of wrong type. The same error is being raised when i use BULK COLLECT.
What should I use to populate l_dat_rec?

Answer 1

In the answer to your previous question, I mentioned that populating the collection would be harder with the %rowtype field. As far as I'm aware, unless you declare an SQL-level object type instead of a record type you can't use bulk collect for this (though it's worth checking if that has changed in 12c perhaps).
I believe you are stuck with using a simpler cursor loop that builds the two fields in your type (i.e. the %rowtype sub-field and the rowid field) separately, and then builds up the collection a line at a time:
create or replace package body dat_pkg is
procedure transform_dat (p_batch_name data_test.batch_name%type)
is
cursor cur_dat is
select rowid, a.*
from data_test a
where batch_name = p_batch_name;
l_dat_tst typ_dat_tst;
l_rec data_test%rowtype;
begin
for rec_dat in cur_dat loop
l_rec.data_id := rec_dat.data_id;
l_rec.data_value := rec_dat.data_value;
l_rec.batch_name := rec_dat.batch_name;
-- or use a counter you increment for this...
l_dat_tst(l_dat_tst.count + 1).data_rec := l_rec;
l_dat_tst(l_dat_tst.count).data_rowid := rec_dat.rowid;
end loop;
-- Do the Transformation here. Example --
for i in 1..l_dat_tst.count loop
if l_dat_tst(i).data_rec.data_value = 'hello' then
l_dat_tst(i).data_rec.data_value := 'was hello';
else
l_dat_tst(i).data_rec.data_value := 'was not hello';
end if;
end loop;
-- update the table
proc_test (p_dat => l_dat_tst);
end transform_dat;
procedure proc_test (p_dat typ_dat_tst)
is
begin
for i in 1..p_dat.count loop
update data_test
set data_value = p_dat(i).data_rec.data_value
where data_id = p_dat(i).data_rec.data_id
and rowid = p_dat(i).data_rowid;
end loop;
end proc_test;
end dat_pkg;
/
As also discussed before, the references to the sub-field-record's fields have to be qualified properly, so I've inserted .data_rec in the references in both procedures. I've changed the dummy transformation to modify the value instead of the ID, as that means no updates were ever going to happen.
Demo with some dummy data:
insert into data_test values (1, 'hello', 'test');
insert into data_test values (2, 'hello', 'test');
insert into data_test values (3, 'hello', 'exclude');
insert into data_test values (4, 'goodbye', 'test');
exec dat_pkg.transform_dat('test');
select * from data_test;
DATA_ID DATA_VALUE BATCH_NAME
---------- -------------------- --------------------
1 was hello test
2 was hello test
3 hello exclude
4 was not hello test