Accept uni or multi-dimensional parameter in PL/SQL function - oracle

Question

I'm trying to write a function nbrJobs() in PL/SQL that counts the number of jobs that a user has had in the past.
In order to do this I first need to determine the "employee id", which can be determined by a pair (first name, last name).
I manage to do this when the arguments of the function are:
nbrJobs(firstName IN VARCHAR2(20), lastName IN VARCHAR2(25))
Then, this simple test runs without any problem:
DECLARE
nbrJobsTotal NUMBER;
BEGIN
SELECT nbrJobs(first_name, last_name) INTO nbrJobsTotal
FROM employees
WHERE employee_id = 101
END
Now, the problem is that my function should also work with that kind of call:
SELECT nbrJobs(first_name, last_name) INTO nbrJobsTotal FROM employees
with table employees containing multiple tuples.
So, now I'm confused about the input parameters type.. Should I use a VARRAY, a nested TABLE, a CURSOR, something else ?
What does a SELECT actually returns if multiple rows are selected?

Answer 1

A PL/SQL function is executed for each row of the SELECT statement. Therefore, if you call your function in a regular SELECT SQL statement you will get a value for each record.
Here is an example by concatenating the first and last name together:
CREATE OR REPLACE FUNCTION NAME (p_FIRST_NAME IN VARCHAR2, p_LAST_NAME IN VARCHAR2)
RETURN VARCHAR2
AS
BEGIN
RETURN p_FIRST_NAME || ' ' || p_LAST_NAME;
END;
/
SELECT first_name, last_name, name(first_name, last_name) FROM HR.employees;
FIRST_NAME LAST_NAME NAME(FIRST_NAME,LAST_NAME)
---------- --------- --------------------------
Ellen Abel Ellen Abel
Sundar Ande Sundar Ande
Mozhe Atkinson Mozhe Atkinson
David Austin David Austin
As you can see, for each row the PL/SQL function is executed and concatenates the first and last name and then is returning the result as a new column to the SELECT statement. There is no need for you to change the function to make it work with multiple row.
Now how would you be executing this inside PL/SQL with the SELECT statement that you have used as example above. You will either have to loop over the results with a cursor or you could use a collection type if you just want to fetch the result of all of the rows into a variable.
Using a cursor (
I'm demonstrating this on my example above by using the Cursor FOR LOOP):
BEGIN
FOR result IN (SELECT first_name, last_name, name(first_name, last_name) name FROM HR.employees) LOOP
DBMS_OUTPUT.PUT_LINE(result.first_name || ', ' || result.last_name || ', ' || result.name);
END LOOP;
END;
/
Ellen, Abel, Ellen Abel
Sundar, Ande, Sundar Ande
Mozhe, Atkinson, Mozhe Atkinson
David, Austin, David Austin
What happens here is that I'm executing the very same SELECT statement but now inside the Cursor FOR LOOP which allows me to loop over each individual row that has been return. In this case I just print the result out into the console.
If you want to just save all the rows into a variable you will have to use a PL/SQL Collection:
DECLARE
-- Specify cursor with expected results
CURSOR c1 IS
SELECT first_name, last_name, name(first_name, last_name) name
FROM HR.employees;
-- Create PL/SQL type for a nested table of the rowtype of the cursor (first_name, last_name, name)
TYPE NameSet IS TABLE OF c1%ROWTYPE;
employees NameSet; -- Instantiate a variable of the nested table of records
BEGIN
-- Assign values to nested table of records:
SELECT first_name, last_name, name(first_name, last_name) name
BULK COLLECT INTO employees
FROM HR.employees;
-- Print nested table of records:
FOR i IN employees.FIRST .. employees.LAST LOOP
DBMS_OUTPUT.PUT_LINE (
employees(i).first_name || ' ' ||
employees(i).last_name || ', ' ||
employees(i).name
);
END LOOP;
END;
/
Ellen Abel, Ellen Abel
Sundar Ande, Sundar Ande
Mozhe Atkinson, Mozhe Atkinson
David Austin, David Austin
As you can see here the same SELECT is executed but here we use BULK COLLECT INTO rather than just INTO. This is because the SELECT is returning more than one row, hence we need to tell the compiler that we do indeed expect that so that the compiler doesn't throw an error that more rows have been returned.
Last but not least, given that you use the variable name nbrJobsTotal in your example SELECT nbrJobs(first_name, last_name) INTO nbrJobsTotal FROM employees above, I think what you really want to try to do here is to sum up all the number of different jobs that employees had in your company. You can accomplish just that but using the built-in SUM() function which is an aggregation function, i.e it will only return one row without a GOUP BY clause:
SELECT SUM(nbrJobs(first_name, last_name)) INTO nbrJobsTotal FROM employees

Answer 2

You can use a Table of Type - to bulk collect all rows returned by the select statement.
Based on your example it would look like this:
DECLARE
nbrJobsTotal NUMBER;
TYPE jobsTotalTable_type IS TABLE OF nbrJobsTotal%TYPE;
jobsTotalTable jobsTotalTable_type ;
BEGIN
--bulk collect results
SELECT nbrJobs(first_name, last_name) BULK COLLECT INTO jobsTotalTable
FROM employees;
--print results
FOR indx IN 1 .. jobsTotalTable.COUNT LOOP
DBMS_OUTPUT.PUT_LINE(jobsTotalTable (indx));
END LOOP;
END