PL/SQL Using SQL%NOTFOUND to raise a user defined exception in a function

Question

I have written the following function for finding the number of items in stock in the item table.
CREATE OR REPLACE function getAmount (ItemID IN NUMBER)
RETURN NUMBER
AS
invalid_id EXCEPTION;
returnedQty number;
BEGIN
Select qty
Into returnedQty
From item
Where itemNo = ItemID;
RETURN (returnedQty);
IF SQL%NOTFOUND THEN
RAISE invalid_id;
END IF;
COMMIT;
Exception
WHEN invalid_id THEN
DBMS_OUTPUT.PUT_LINE('Invalid ID entered');
END getAmount;
The function compiles successfully, although there is a problem that Oracle is not handling my user-defined exception invalid_id
If I use the following for a valid itemID:
DECLARE
return_value number;
BEGIN
return_value := getAmount(1);
DBMS_OUTPUT.PUT_LINE (return_value);
END;
then the function returns the quantity of items in stock correctly.
However, if I enter an incorrect itemID, say 20
DECLARE
return_value number;
BEGIN
return_value := getAmount(20);
DBMS_OUTPUT.PUT_LINE (return_value);
END;
The invalid_id exception is not raised, and the Oracle errors says: no_data_found and the function has not returned a value. If I add a no_data_found exception, this works perfectly, but for this assignment I must write my own user-defined error.
Any help would be very much appreciated!
Thank you.

Answer 1

What you're trying to do, is to use an implicit cursor. Implicit cursors will raise no_data_found and too_many_rows in case of an error. Explicit cursors will not.
Also, you have a small coding error ... a little rewrite and your code will do as you want to:
CREATE OR REPLACE function getAmount (ItemID IN NUMBER)
RETURN NUMBER
AS
invalid_id EXCEPTION;
returnedQty number;
cursor citem is
Select qty
From item
Where itemNo = ItemID;
fnd boolean;
BEGIN
open citem;
fetch citem Into returnedQty;
fnd := citem%found;
close citem;
if fnd then
RETURN (returnedQty);
else
RAISE invalid_id;
END IF;
/* Do not catch your own exception or the calling code will not receive it.
Exception
WHEN invalid_id THEN
DBMS_OUTPUT.PUT_LINE('Invalid ID entered');
END getAmount;
Your problem is, you're not using a package. This means, that the calling code has no way of catching this user-defined exeption. The correct way is to define a package, define the userdefined exception in the package and refer to it in your code:
create or replace package amt
is
invalid_id EXCEPTION;
function getAmount (ItemID IN NUMBER)
RETURN NUMBER;
end;
create or replace package body amt
is
function getAmount (ItemID IN NUMBER)
RETURN NUMBER
AS
returnedQty number;
cursor citem is
Select qty
From item
Where itemNo = ItemID;
fnd boolean;
BEGIN
open citem;
fetch citem Into returnedQty;
fnd := citem%found;
close citem;
if fnd then
RETURN (returnedQty);
else
RAISE invalid_id;
END IF;
end;
end; -- end package
To do a simple test, you'll do:
declare
a number;
begin
a := amt.getAmount(123);
dbms_output.put_line('The result is '||a);
exception
when amt.invalid_id then
dbms_output.put_line('ID not found');
end;
Other ways is to raise the exception in your no_data_found block.
But I'm REALLY puzzled if there's really someone out there who prefers a user-defined exception instead of the clearly defined NO_DATA_FOUND exception.
I can't be sure of course, but are you sure what they mean is they don't want ANY exception? That's a pretty common requirement.
In general - explicit cursors are a bit faster because Oracle does not have to do a second fetch to determine TOO_MANY_ROWS. And it's not too much additional writing. Explicit cursors will never raise exceptions - unless you use dynamic sql and your SQL is invalid.
Good luck