Are there alternative methods for saying 'next' in a pl/sql for loop? - oracle

Question

So I've got a for loop that processes a list of IDs and has some fairly complex things to do. Without going into all the ugly details, basically this:
DECLARE
l_selected APEX_APPLICATION_GLOBAL.VC_ARR2;
...snip...
BEGIN
-- get the list ids
l_selected := APEX_UTIL.STRING_TO_TABLE(:P4_SELECT_LIST);
-- process each in a nice loop
FOR i IN 1..l_selected.count
LOOP
-- do some data checking stuff...
-- here we will look for duplicate entries, so we can noop if duplicate is found
BEGIN
SELECT county_id INTO v_dup_check FROM org_county_accountable
WHERE organization_id = :P4_ID AND county_id = v_county_id;
-- NEXT;! NOOP;! but there is no next!
EXCEPTION WHEN NO_DATA_FOUND THEN
dbms_output.put_line('no dups found, proceeding');
END;
-- here we have code we only want to execute if there are no dupes already
IF v_dup_check IS NULL THEN
-- if not a duplicate record, proceed...
ELSE
-- reset duplicate check variable
v_dup_check := NULL;
END;
END LOOP;
END;
How I normally handle this is by selecting into a value, and then wrap the following code in an IF statement checking to make sure that duplicate check variable is NULL. But it's annoying. I just want to be able to say NEXT; or NOOP; or something. Especially since I already have to catch the NO_DATA_FOUND exception. I suppose I could write a letter to Oracle, but I'm curious how others handle this.
I could also wrap this in a function, too, but I was looking for something a little cleaner/simpler.

Answer 1

Oracle 11g adds a C-style "continue" loop construct to PL/SQL, which syntactically sounds like what you're looking for.
For your purposes, why not just eliminate the duplicates prior to entering the loop? This could be done by querying l_selected using a table function, and then filtering out records you don't want instead of iterating over every value. Something like...
declare
l_selected APEX_APPLICATION_GLOBAL.VC_ARR2;
cursor no_dups_cur (p_selected APEX_APPLICATION_GLOBAL.VC_ARR2) is
select * from (
select selected.*,
count(*) over (partition by county_id) cnt -- analytic to find counts grouped by county_id
from table(p_selected) selected -- use table function to treat VC_ARR2 like a table
) where cnt = 1 -- remove records that have duplicate county_ids
;
begin
l_selected := APEX_UTIL.STRING_TO_TABLE(:P4_SELECT_LIST);
for i in no_dups_cur(l_selected) loop
null; -- do whatever to non-duplicates
end loop;
end;
Just substitute the logic for determining a "duplicate" with your own (didn't have enough info from your example to really answer that part)

Answer 2

Instead of catching NO_DATA_FOUND, how about SELECTing the number of matching entries into a variable, say l_count, and proceeding if this count works out to be zero? Something like the following:
DECLARE
l_selected APEX_APPLICATION_GLOBAL.VC_ARR2;
l_count INTEGER;
...snip...
BEGIN
-- get the list ids
l_selected := APEX_UTIL.STRING_TO_TABLE(:P4_SELECT_LIST);
-- process each in a nice loop
FOR i IN 1..l_selected.count
LOOP
-- do some data checking stuff...
-- here we will count duplicate entries, so we can noop if duplicate is found
SELECT COUNT(*) INTO l_count FROM org_county_accountable
WHERE organization_id = :P4_ID AND county_id = v_county_id;
IF l_count = 0 THEN
-- here we have code we only want to execute if there are no dupes already
-- if not a duplicate record, proceed...
END IF;
END LOOP;
END;

Answer 3

To count the number of rows is also possible (see Pourquoi Litytestdata) but you can also do what you want to do in the when_no_data_found exception block.
declare
l_selected apex_application_global.vc_arr2;
l_county_id org_county_accountable.count_id%type;
begin
l_selected := apex_util.string_to_table(:p4_select_lst);
for i in l_selected.first..l_selected.last loop
begin
select count_id
into l_county_id
from org_county_accountable
where organization_id = :p4_id
and county_id = v_county_id;
exception
when no_data_found then
-- here we have code we only want to execute if there are no dupes already
-- if not a duplicate record, proceed...
end;
end loop;
end;

Answer 4

<xmp>
<<next_loop>>
loop
...
...
if ....
then
goto next_loop;
</xmp>

Answer 5

This is a case where a GOTO statement might be useful. See the Oracle Documentation in the control structures to see how to do this. Also, you may want to search around here to find out how to query for the existence of a record. Running a query and waiting for an exception isn't optimal.

Answer 6

Another way - turn the check into a local function:
DECLARE
l_selected APEX_APPLICATION_GLOBAL.VC_ARR2;
...snip...
FUNCTION dup_exists
( p_org_id org_county_accountable.organization_id%TYPE
, p_county_id org_county_accountable.county_id%TYPE
) RETURN BOOLEAN
IS
v_dup_check org_county_accountable.county_id%TYPE;
BEGIN
SELECT county_id INTO v_dup_check FROM org_county_accountable
WHERE organization_id = p_org_id AND county_id = p_county_id;
RETURN TRUE;
EXCEPTION WHEN NO_DATA_FOUND THEN
RETURN FALSE;
END;
BEGIN
-- get the list ids
l_selected := APEX_UTIL.STRING_TO_TABLE(:P4_SELECT_LIST);
-- process each in a nice loop
FOR i IN 1..l_selected.count
LOOP
-- do some data checking stuff...
-- here we have code we only want to execute if there are no dupes already
IF NOT dup_exists (:P4_ID, v_county_id) THEN
-- if not a duplicate record, proceed...
END;
END LOOP;
END;
Of course, the local function could be re-written to use the count method if you prefer:
FUNCTION dup_exists
( p_org_id org_county_accountable.organization_id%TYPE
, p_county_id org_county_accountable.county_id%TYPE
) RETURN BOOLEAN
IS
l_count INTEGER;
BEGIN
SELECT COUNT(*) INTO l_count
FROM org_county_accountable
WHERE organization_id = p_org_id AND county_id = p_county_id;
RETURN (l_count > 0);
END;

Answer 7

Another method is to raise and handle a user-defined exception:
DECLARE
l_selected APEX_APPLICATION_GLOBAL.VC_ARR2;
duplicate_org_county EXCEPTION;
...snip...
BEGIN
-- get the list ids
l_selected := APEX_UTIL.STRING_TO_TABLE(:P4_SELECT_LIST);
-- process each in a nice loop
FOR i IN 1..l_selected.count
LOOP
BEGIN
-- do some data checking stuff...
-- here we will look for duplicate entries, so we can noop if duplicate is found
BEGIN
SELECT county_id INTO v_dup_check FROM org_county_accountable
WHERE organization_id = :P4_ID AND county_id = v_county_id;
RAISE duplicate_org_county;
EXCEPTION WHEN NO_DATA_FOUND THEN
dbms_output.put_line('no dups found, proceeding');
END;
-- here we have code we only want to execute if there are no dupes already
EXCEPTION
WHEN duplicate_org_county THEN NULL;
END;
END LOOP;
END;
I wouldn't normally do this, but if there were half a dozen reasons to jump to the next record, this might be preferable to multiple nested IFs.

Answer 8

I know this is an oldie but I couldn't help notice that none of the answers above take into account the cursor attributes:
There are four attributes associated with cursors: ISOPEN, FOUND, NOTFOUND, and ROWCOUNT. These attributes can be accessed with the % delimiter to obtain information about the state of the cursor.
The syntax for a cursor attribute is:
cursor_name%attribute
where cursor_name is the name of the explicit cursor.
So in this case you could use ROWCOUNT (which indicates the number of rows fetched so far) for your purposes, like this:
declare
aux number(10) := 0;
CURSOR cursor_name is select * from table where something;
begin
select count(*) into aux from table where something;
FOR row IN cursor_name LOOP
IF(aux > cursor_name%ROWCOUNT) THEN 'do something is not over';
ELSE 'do something else';
END IF;
END LOOP;
end;