Stored Procedures in PL/SQL
Stored Procedures in PL/SQL
Many modern databases support a more procedural approach to databases—they allow you
to write procedural code to work with data. Usually, it takes the form of SQL interweaved
with the more familiar IF statements, etc.
Note that this has nothing to do with accessing the database. You can access any database
from virtually any language. What we’re talking about is the code that is executed by the
database server.
While there are many various ‘database’ languages, we will only talk about the primary
two: T-SQL, which is supported by SQL Server and Sybase, and PL/SQL, which is supported
by Oracle.
Many other languages may be supported. For example, Oracle allows you to write stored
procedures and triggers in Java, etc.
2 PL/SQL
Besides plain vanilla SQL, Oracle supports PL/SQL. The PL stands for Procedural Lan-
guage, which means you can have things like IF statements, loops, variables, and other
procedural things along with declarative SQL statements. PL/SQL
2.1 Variables
Just as most procedural languages, PL/SQL has some sort of variables. The types of variables
are plain SQL column types that you’re all used to. You can also refer to a type of a particular
column explicitly by specifying the fully qualified column name (tablename.columname)
followed by %TYPE. For example: PRODUCT.PRICE%TYPE.
Similarly, we can refer to a particular row as a single type. Again, you declare it by
referring to a database table directly: PRODUCT%ROWTYPE. This would refer to a single record
stored in the PRODUCT table.
Along with the above mentioned, some common types are: BOOLEAN, DATE, NUMBER, CHAR,
and VARCHAR2.
We declare variables of these types similarly to specifying columns in tables. First, we
list the name of the variable, then the type we want it to have. For example, to declare a
price variable of a fixed point NUMBER, we might do something like this:
PRICE NUMBER(6,2);
DECLARE
variable_declarations
BEGIN
procedural_code
1
EXCEPTION
error_handling
END;
2.2.1 Declare
The declare part is where variable declaration goes. All used variables must be declared in
this section. This is also the place where other more exotic variable types are declared, like
cursors and exceptions.
2.2.2 Begin
This is the part we’re most interested in. This is where the bulk of your programs shall be
placed. Here, you can have IF statements, loops, etc.
2.2.3 Exceptions
The exception section is where we place error handling code. We will talk about it more
depth in subsequent lessons.
2.2.4 End
The end signifies the end of this program block.
2.3 Operators
PL/SQL supports several operators to do various things. Table 1 lists some of the more
common operators.
SET SERVEROUTPUT ON
What this does is enable you to view output in SQL*Plus window whenever your pro-
grams attempts to write some output to the screen.
Now, let’s get on with the show. Type the following into the SQL*Plus window as is:
BEGIN
DBMS_OUTPUT.PUT_LINE(’Hello World’);
END;
You’ll notice that it doesn’t run as an average SQL statement. To make it run, you must
type the ’/’ character on a line by itself. And you’ll notice:
2
SQL> BEGIN
2 DBMS_OUTPUT.PUT_LINE(’Hello World’);
3 END;
4 /
Hello World
SQL> BEGIN
2 DBMS_OUTPUT.PUT_LINE(’The time now is: ’);
3 DBMS_OUTPUT.PUT_LINE(SYSDATE);
4 END;
5 /
The time now is:
31-JUL-02
We’re not done with this simple example yet. We can also modify the DATE format:
∗∗ Exponentiation
∗ Multiplication
/ Division
+ Addition
− Subtraction
− Negation
:= Assignment
= Equals Comparison
<> Not Equals Comparison
!= Not Equals Comparison
> Greater Than Comparison
< Less Than Comparison
>= Greater Than or Equal Comparison
<= Less Than or Equal Comparison
AN D The obvious AND operation
OR The obvious OR operation
:= Assignment
|| String Concatenation
3
SQL> BEGIN
2 DBMS_OUTPUT.PUT_LINE(’The time now is: ’);
3 DBMS_OUTPUT.PUT_LINE(TO_CHAR(SYSDATE,’MM/DD/YYYY’));
4 END;
5 /
The time now is:
07/31/2002
SQL> BEGIN
2 DBMS_OUTPUT.PUT_LINE(’The time now is: ’ || SYSDATE);
3 END;
4 /
The time now is: 31-JUL-02
Note that || was used to concatenate the string ’The time is now: ’ with the
SYSDATE. Some of the more useful PL/SQL are listed in Table 3.
4
SQL> SELECT SUBSTR(’HELLO’,2,4) FROM DUAL;
SUBS
----
ELLO
IF condition THEN
program_statements
END IF;
Assuming we all know how to program, and know what IF statements are, I’m not going
to spend too much time on the obvious.
An example program that uses an IF statement is:
DECLARE
A NUMBER(6);
B NUMBER(6);
BEGIN
A := 23;
B := A * 5;
IF A < B THEN
DBMS_OUTPUT.PUT_LINE(’Ans: ’ || A || ’ is less than ’ || B);
END IF;
END;
5
IF condition THEN
if_condition_is_true_code
ELSE
if_condition_is_false_code
END IF;
DECLARE
A NUMBER(6);
B NUMBER(6);
BEGIN
A := 23;
B := A / 5;
IF A < B THEN
DBMS_OUTPUT.PUT_LINE(’Ans: ’ || A || ’ is less than ’ || B);
ELSE
DBMS_OUTPUT.PUT_LINE(’Ans: ’ || A || ’ is greater than ’ || B);
END IF;
END;
Note that we’ve also modified the B := A * 5 to B := A / 5 in order to test the ELSE
condition.
DECLARE
A NUMBER(6);
B NUMBER(6);
C NUMBER(6);
ABCMAX NUMBER(6);
BEGIN
A := 23;
B := A / 5;
C := B * 7;
IF A > B THEN
IF A > C THEN
ABCMAX := A;
ELSE
ABCMAX := C;
END IF;
ELSE
IF B > C THEN
6
ABCMAX := B;
ELSE
ABCMAX := C;
END IF;
END IF;
DBMS_OUTPUT.PUT_LINE(’Max of: ’ || A || ’, ’ || B ||
’, and ’ || C || ’ is ’ || ABCMAX);
END;
The code above finds the maximum value (ABCMAX) of three variables (A, B, and C).
The code looks self explanatory; so we won’t spend much time on it.
DECLARE
NGRADE NUMBER;
LGRADE CHAR(2);
BEGIN
NGRADE := 82.5;
7
DBMS_OUTPUT.PUT_LINE(’Grade ’ || NGRADE || ’ is ’ || LGRADE);
END;
Which for our particular example number grade produces output:
Grade 82.5 is B
8
2.13 LOOP ... EXIT Loop
The general format of such a loop is:
LOOP
various_statements
IF condition THEN
EXIT;
END IF;
various_statements
END LOOP;
This loop is very similar to an infinite loop in C/C++, where you use break; to terminate
the loop; in this case, the EXIT; command takes the form of break.
Note that we can place various program statements before the exiting IF statement and
after, which gives us great flexibility about when and how the condition is evaluated.
An example of such a loop would be:
DECLARE
I NUMBER(6);
BEGIN
I := 1;
LOOP
DBMS_OUTPUT.PUT_LINE(’aI: ’ || I);
I := I + 1;
IF I > 5 THEN
EXIT;
END IF;
DBMS_OUTPUT.PUT_LINE(’bI: ’ || I);
END LOOP;
END;
aI: 1
bI: 2
aI: 2
bI: 3
aI: 3
bI: 4
aI: 4
bI: 5
aI: 5
9
Note that you should SET SERVEROUTPUT ON; in order to see the output in SQL*Plus
screen.
Also, it would be very helpful if you trace the above program to ensure that you under-
stand how the loop functions and why the results look as they do. I shall not provide the
output for the following code, and expect you to run it yourself.
LOOP
various_statements
EXIT WHEN condition;
various_statements
END LOOP;
DECLARE
I NUMBER(6);
BEGIN
I := 1;
LOOP
DBMS_OUTPUT.PUT_LINE(’aI: ’ || I);
I := I + 1;
EXIT WHEN I > 5;
DBMS_OUTPUT.PUT_LINE(’bI: ’ || I);
END LOOP;
END;
You should run this code yourself. It would actually be more helpful if you write out the
output first, and then compare it to the actual results.
WHILE condition
LOOP
various_statements
END LOOP;
10
Our typical (as in typical for these class notes) would be:
DECLARE
I NUMBER(6);
BEGIN
I := 1;
WHILE I <= 5
LOOP
DBMS_OUTPUT.PUT_LINE(’aI: ’ || I);
I := I + 1;
DBMS_OUTPUT.PUT_LINE(’bI: ’ || I);
END LOOP;
END;
Just as with the previous code, you should try to figure out what the output is, and then
run it to see if your trace was correct. Tracing questions such as these are fair game for
quizzes and tests.
The start and end values must be integers, and are always incremented by one. An
example of such a loop would be:
BEGIN
FOR I IN 1..5
LOOP
DBMS_OUTPUT.PUT_LINE(’I: ’ || I);
END LOOP;
END;
Notice that we never actually directly initialize I, or even declare it! It is done implicitly
for us by the FOR loop. You should run this code to ensure you understand it.
You can also use other variables to loop on. For example, to loop from J to K, you’d do
something like:
DECLARE
J NUMBER(6);
K NUMBER(6);
11
BEGIN
J := 7;
K := 2;
FOR I IN K..J
LOOP
DBMS_OUTPUT.PUT_LINE(’I: ’ || I);
END LOOP;
END;
Again, notice that we never actually initialize nor declare I. In fact, the I in the loop is
a totally different variable. Even if you have an I variable declared, the loop will still use its
own version. You can verify that by running this code:
DECLARE
I NUMBER(6);
BEGIN
I := 7;
DBMS_OUTPUT.PUT_LINE(’BEFORE LOOP I: ’ || I);
FOR I IN 1..5
LOOP
DBMS_OUTPUT.PUT_LINE(’IN LOOP I: ’ || I);
END LOOP;
DBMS_OUTPUT.PUT_LINE(’AFTER LOOP I: ’ || I);
END;
BEFORE LOOP I: 7
IN LOOP I: 1
IN LOOP I: 2
IN LOOP I: 3
IN LOOP I: 4
IN LOOP I: 5
AFTER LOOP I: 7
Which illustrates that the value of our declared variable I is unchanged by the loop (and
that the loop internally has I declared which is different from our explicitly declared I).
2.17 Cursors
Before we move on with our discussion of the next and last loop construct, we must cover
the concept of Cursors.
Oracle has two major different types of cursors. One is implicit and the other one is
explicit.
12
2.18 Implicit Cursor
Implicit cursors can be generated every time you do a SELECT statement in PL/SQL. The
general format goes something like this:
SELECT selectfields INTO declared_variables FROM table_list WHERE search_criteria;
The only catch is that the search criteria must return one and only one result. If it
returns zero, or more than one, an error is generated.
For example, lets say we wanted to get the name and price of some specific product
(identified by PRODUCT_ID):
DECLARE
NAME PRODUCT.DESCRIPTION%TYPE;
AMOUNT PRODUCT.PRICE%TYPE;
BEGIN
SELECT DESCRIPTION,PRICE INTO NAME, AMOUNT
FROM PRODUCT WHERE PRODUCT_ID = 4;
DBMS_OUTPUT.PUT_LINE(’PRICE OF ’ || NAME || ’ IS ’ || AMOUNT);
END;
Which faithfully displays out:
PRICE OF keyboard IS 19.95
Assuming the “keyboard” is in the database and has PRODUCT ID = 4 (and has that
price).
Note that we used the table’s types, which brings up another issue: Now is a pretty good
time to illustrate the ROWTYPE type. Let’s rewrite the above using that.
DECLARE
P PRODUCT%ROWTYPE;
BEGIN
SELECT * INTO P FROM PRODUCT WHERE PRODUCT_ID = 4;
DBMS_OUTPUT.PUT_LINE(’PRICE OF ’ || P.DESCRIPTION || ’ IS ’ || P.PRICE);
END;
Notice that the code got a lot smaller since we don’t have to worry about defining every
single variable for retrieval purposes. We retrieve a whole row of data at a time. The output
of the above code is exactly the same as the previous.
13
3 Opening an Explicit Cursor
In order to use an explicit cursor, you must open it. You do that with a simple:
OPEN cursorname;
(obviously you have to do that inside the code section, between BEGIN and END).
We shall do some examples when we learn our cursor loops, so hang on...
CLOSE cursorname;
Cursors will be closed automatically once your code exits, but it’s still a good idea to
close them explicitly.
DECLARE
P PRODUCT%ROWTYPE;
CURSOR PRODUCTCURSOR IS
SELECT * FROM PRODUCT;
BEGIN
OPEN PRODUCTCURSOR;
LOOP
FETCH PRODUCTCURSOR INTO P;
EXIT WHEN PRODUCTCURSOR%NOTFOUND;
DBMS_OUTPUT.PUT_LINE(’PRICE OF ’ || P.DESCRIPTION || ’ IS ’ || P.PRICE);
END LOOP;
CLOSE PRODUCTCURSOR;
END;
14
Go through the code line by line. First, we declare our P variable which is a ROWTYPE
from table PRODUCT. We then declare our CURSOR, which simply selects everything from the
PRODUCT table.
Our code then proceeds to OPEN the cursor. We then fall into our standard loop (which
we learned about earlier), and FETCH results from the CURSOR. We EXIT the loop if we got
no more results (the PRODUCTCURSOR%NOTFOUND condition). If we did not exit the loop, we
output product description and product price.
In the end, we just CLOSE the cursor. Depending on what you have in your PRODUCT
table, the results of the code may look similar to this:
You should go through the code, trace it, run it, and make sure you understand it.
Let us rewrite our example program (presented earlier) to use this new type of loop:
15
DECLARE
P PRODUCT%ROWTYPE;
CURSOR PRODUCTCURSOR IS
SELECT * FROM PRODUCT;
BEGIN
FOR P IN PRODUCTCURSOR LOOP
DBMS_OUTPUT.PUT_LINE(’PRICE OF ’ || P.DESCRIPTION || ’ IS ’ || P.PRICE);
END LOOP;
END;
Notice that the code got quite a bit simpler, with lots of cursor handling code gone; which
is now being handled by the loop itself.
If you’re really into optimization, you might want to improve the above code not to return
the whole %ROWTYPE but individual fields which we’re displaying, for example:
DECLARE
CURSOR PRODUCTCURSOR IS
SELECT DESCRIPTION,PRICE FROM PRODUCT;
BEGIN
FOR P IN PRODUCTCURSOR LOOP
DBMS_OUTPUT.PUT_LINE(’PRICE OF ’ || P.DESCRIPTION || ’ IS ’ || P.PRICE);
END LOOP;
END;
Notice several things about the code: that we no longer declare P which is used for
loop purposes. Also notice that our cursor is no longer returning everything, but just two
individual fields which we’re displaying.
From that point on, the user someusername will be allowed to create, drop, and replace
procedures and functions.
16
3.7 PROCEDURES
Procedures are code fragments that don’t normally return a value, but may have some outside
effects (like updating tables). The general format of a procedure is:
PROCEDURE procedure_name IS
BEGIN
procedure_body
END;
Of course, you’ll usually be either creating or replacing the procedure, so you’d want to
add on CREATE (OR REPLACE) to the declaration. For example, to create (or replace) a HELLO
procedure, you might do something like this:
CREATE OR REPLACE
PROCEDURE HELLO IS
BEGIN
DBMS_OUTPUT.PUT_LINE(’Hello World’);
END;
The above declares a HELLO procedure that just displays ’Hello World’. You can run
it as part of a code fragment, or inside other procedures (or functions). For example:
BEGIN
HELLO();
END;
CALL HELLO();
CREATE OR REPLACE
PROCEDURE procedure_name ( parameters ) IS
BEGIN
procedure_body
END;
Where procedure_name can be any valid SQL name, parameters is a list of parameters to
this procedure (we’ll discuss them later), and procedure_body is various PL/SQL statements
that make up the logic of the procedure.
17
3.8.1 Parameters
The parameters (or arguments) are optional. You don’t have to specify anything (not even
the parenthesis). For example, a sample procedure, which you no doubt have already seen:
CREATE OR REPLACE
PROCEDURE HELLOWORLD IS
BEGIN
DBMS_OUTPUT.PUT_LINE(’Hello World!’);
END;
Never actually defines any parameters. What’s the use of a procedure that doesn’t take
any parameters and doesn’t return anything? Well, you may be interested in the procedure’s
side effects, like in our case, we’re interested in our procedure displaying ’Hello World!’ and
nothing else. There may be many instances where you may want to just do something to
the database, without any particular parameters, and without returning anything.
Anyway, this section is about parameters so let’s talk about parameters.
Parameters are defined in a similar way as in a CREATE TABLE statement, which is similar
to how variables are declared. You first specify the name of the variable, and then the type.
For example:
(N INT)
Would setup some procedure to accept an INT variable named N. Writing a simple pro-
cedure to display a variable name, you can come up with something like this:
CREATE OR REPLACE
PROCEDURE DISPN (N INT) IS
BEGIN
DBMS_OUTPUT.PUT_LINE(’N is ’ || N);
END;
You can also have multiple parameters. For example, you can accept A and B and display
their sum and product.
CREATE OR REPLACE
PROCEDURE DISP_AB (A INT, B INT) IS
BEGIN
DBMS_OUTPUT.PUT_LINE(’A + B = ’ || (A + B));
DBMS_OUTPUT.PUT_LINE(’A * B = ’ || (A * B));
END;
18
Which when ran, displays something like (depending on the values you provide):
Btw, it should be noted that you can use any PL/SQL type as an argument. For example,
VARCHAR and others are perfectly acceptable. For example:
CREATE OR REPLACE
PROCEDURE DISP_NAME (NAME VARCHAR) IS
BEGIN
DBMS_OUTPUT.PUT_LINE(’Hi ’ || NAME || ’!’);
END;
CREATE OR REPLACE
PROCEDURE SUM_AB (A INT, B INT, C OUT INT) IS
BEGIN
C := A + B;
END;
Notice that the above code does not display the resulting sum, it just changes the value
of the C parameter. Also notice the word OUT right after the declaration of C parameter
name.
Anyway, we will use a code fragment to call the procedure:
DECLARE
R INT;
BEGIN
SUM_AB(23,29,R);
DBMS_OUTPUT.PUT_LINE(’SUM IS: ’ || R);
END;
19
SUM IS: 52
Notice how we called the procedure with an argument to eventually retrieve the OUT
result.
There is also the other special way of passing parameters: IN OUT. What that means is
that we first can read the parameter, then we can change it. For example, we can write a
procedure that doubles a number:
CREATE OR REPLACE
PROCEDURE DOUBLEN (N IN OUT INT) IS
BEGIN
N := N * 2;
END;
DECLARE
R INT;
BEGIN
R := 7;
DBMS_OUTPUT.PUT_LINE(’BEFORE CALL R IS: ’ || R);
DOUBLEN(R);
DBMS_OUTPUT.PUT_LINE(’AFTER CALL R IS: ’ || R);
END;
Notice how this particular call first grabbed the value of a parameter, then set it in order
to return the double of the value.
You can generally intermix these various ways of passing parameters (along with various
types). You can use these to setup return values from procedures, etc.
That’s all there is to stored procedures. We will do some practice exercises and more
experimentation, but overall, that’s all there is to them.
20
3.11 Functions
Functions are special types of procedures that have the capability to return a value.
It is a very shady question of when to use what, either functions or procedures. A good
rule of thumb is: if you’re interested in the “results” of the code, then you use a function,
and return those results. If you’re interested in the “side effects” (like table updates, etc.)
and not about the ”result” when you should use a procedure. Usually it doesn’t affect your
code all that much if you use a procedure or a function.
CREATE OR REPLACE
FUNCTION function_name (function_params) RETURN return_type IS
BEGIN
function_body
RETURN something_of_return_type;
END;
For example, to write a function that computes the sum of two numbers, you might do
something like this:
CREATE OR REPLACE
FUNCTION ADD_TWO (A INT,B INT) RETURN INT IS
BEGIN
RETURN (A + B);
END;
To run it, we’ll write a small piece of code that calls this:
BEGIN
DBMS_OUTPUT.PUT_LINE(’RESULT IS: ’ || ADD_TWO(12,34));
END;
RESULT IS: 46
All of a sudden, we know how to make functions (since we already know how to crate
procedures). That’s really all there is to it.
21
3.13 Dropping Functions
To drop a function, you do it in a similar way to a procedure. You simply say:
Oh, btw, to display the list of procedures/functions or plain general user objects that
you have you can run a query:
SELECT OBJECT_NAME
FROM USER_OBJECTS
WHERE OBJECT_TYPE = ’FUNCTION’;
22