PL/SQL

Oracle’s Version of Triggers

1
 PL/SQL
◆ Oracle uses a variant of SQL/PSM
which it calls PL/SQL.
◆ PL/SQL not only allows you to create
and store procedures or functions,
but it can be run from the generic
query interface (sqlplus), like any
SQL statement.
◆ Triggers are a part of PL/SQL.
2
 Trigger Differences
◆ Compared with SQL standard
triggers, Oracle has the following
differences:
1. Action is a PL/SQL statement.
2. New/old tuples referenced
automatically.
3. Strong constraints on trigger actions
designed to make certain you can’t
fire off an infinite sequence of
3
triggers.
 Order of Oracle Trigger
Elements
1. CREATE TRIGGER
2. Event, e.g., AFTER INSERT …
3. FOR EACH ROW, if desired.
4. Condition.
5. Action.
6. A dot and the word “run”. These
cause the trigger to be installed in
the database. 4
 New/Old Tuples
◆ Instead of a REFERENCING clause,
Oracle assumes that new tuples
are referred to as “new” and old
tuples by “old.”
◆ Also, for statement-level triggers:
“newtable” and “oldtable”.
◆ In actions, but not in conditions,
you must prefix “new,” etc., by a
colon. 5
 Example: BeerTrig
◆ Recall our example BeerTrig, which
inserted a beer name into Beers
whenever a tuple was inserted into
Sells with a beer that was not
mentioned in Beers.
◆ Here’s the Oracle version of that
same trigger.

6
 BeerTrig in Oracle SQL
CREATE OR REPLACE TRIGGER BeerTrig
AFTER INSERT ON Sells
FOR EACH ROW
WHEN (new.beer NOT IN
(SELECT name FROM Beers))
BEGIN
INSERT INTO BEERS(name) VALUES(:new.beer);
END; Needed to store
. trigger as an Notice “new” is understood.
element of the Also, colon used only in
run the action.
database 7
 Another Example
◆ Recall PriceTrig, which stores in
the relation Ripoffbars(bar) the
name of any bar that raises the
price of any beer by more than $1.
◆ Here’s the Oracle version.

8
 PriceTrig in Oracle
CREATE OR REPLACE TRIGGER PriceTrig
AFTER UPDATE OF price ON Sells
FOR EACH ROW
WHEN (new.price > old.price + 1.00)
BEGIN
INSERT INTO RipoffBars
VALUES(:new.bar);
END;
.
run 9
 Oracle Limitation on
Relations Affected
◆ Each trigger is on some one
relation R, mentioned in the event.
◆ The SQL standard puts no
constraint on which relations,
including R, can be modified in the
action.
◆ As a result, infinite sequences of
triggered events are possible.
10
 Example: Infinite
Triggering
◆ Let R(x) be a unary relation that is
a set of integers.
◆ Easy to write a trigger with event
INSERT ON R, that as action,
inserts i +1 if i was the integer
that awakened the trigger.
◆ Results in a never-ending
sequence of inserts.
11
 Oracle Limitation
◆ Oracle is overly conservative about
what relations can be changed
when the event is on R.
◆ R surely must not be subject to
any modification in the action.
◆ But much trickier: any relation that
is linked to R by a chain of foreign-
key constraints may not be
changed either.
12
 Example: Foreign-Key
Chains
◆ Suppose R.a is a foreign key,
referencing S.b.
◆ Also, T.c is a foreign key
referencing S.b.
◆ Then in a trigger on relation R,
neither T nor S may be modified.

13
 PL/SQL
◆ In addition to stored procedures,
one can write a PL/SQL statement
that looks like the body of a
procedure, but is executed once,
like any SQL statement typed to
the generic interface.
◗ Oracle calls the generic interface
“sqlplus.”
◗ PL/SQL is really the “plus.” 14
 Form of PL/SQL
Statements
DECLARE
<declarations>
BEGIN
<statements>
END;
.
run
◆ The DECLARE section is optional.
15
Form of PL/SQL Procedure
CREATE OR REPLACE PROCEDURE
<name> (<arguments>) AS Notice AS
needed here
<optional declarations>
BEGIN
<PL/SQL statements>
END;
. Needed to store
procedure in database
run
16
 PL/SQL Declarations and
Assignments
◆ The word DECLARE does not
appear in front of each local
declaration.
◗ Just use the variable name and its
type.
◆ There is no word SET in
assignments, and := is used in
place of =.
◗ Example: x := y; 17
 PL/SQL Procedure
Parameters
◆ There are several differences in
the forms of PL/SQL argument or
local-variable declarations,
compared with the SQL/PSM
standard:
1. Order is name-mode-type, not
mode-name-type.
2. INOUT is replaced by IN OUT in
PL/SQL.
18
3. Several new types.
 PL/SQL Types
◆ In addition to the SQL types,
NUMBER can be used to mean INT or
REAL, as appropriate.
◆ You can refer to the type of attribute
x of relation R by R.x%TYPE.
◗ Useful to avoid type mismatches.
◗ Also, R%ROWTYPE is a tuple whose
components have the types of R’s
attributes. 19
 Example:JoeMenu
◆ Recall the procedure JoeMenu(b,p)
that adds beer b at price p to the
beers sold by Joe (in relation Sells).
◆ Here is the PL/SQL version.

20
 Procedure JoeMenu in
PL/SQL
CREATE OR REPLACE PROCEDURE JoeMenu (
b IN Sells.beer%TYPE,
p IN Sells.price%TYPE Notice these types
) AS will be suitable
for the intended
BEGIN uses of b and p.
INSERT INTO Sells
VALUES (’Joe’’s Bar’, b, p);
END;
.
run 21
 PL/SQL Branching
Statements
◆ Like IF … in SQL/PSM, but:
◆ Use ELSIF in place of ELSEIF.
◆ Viz.: IF … THEN … ELSIF … ELSIF …
ELSE … END IF;

22
 PL/SQL Loops
◆ LOOP … END LOOP as in SQL/PSM.
◆ Instead of LEAVE … , PL/SQL uses
EXIT WHEN <condition>
◆ And when the condition is that
cursor c has found no tuple, we
can write c%NOTFOUND as the
condition.

23
 PL/SQL Cursors
◆ The form of a PL/SQL cursor
declaration is:
CURSOR <name> IS <query>;
◆ To fetch from cursor c, say:
FETCH c INTO <variable(s)>;

24
Example: JoeGouge() in PL/SQL
◆ Recall JoeGouge() sends a cursor
through the Joe’s-Bar portion of
Sells, and raises by $1 the price of
each beer Joe’s Bar sells, if that
price was initially under $3.

25
 Example: JoeGouge()
Declarations
CREATE OR REPLACE PROCEDURE
JoeGouge() AS
theBeer Sells.beer%TYPE;
thePrice Sells.price%TYPE;
CURSOR c IS
SELECT beer, price FROM Sells
WHERE bar = ’Joe’’s Bar’;

26
 Example: JoeGouge Body
BEGIN
OPEN c;
LOOP How PL/SQL
breaks a cursor
FETCH c INTO theBeer, thePrice;loop
EXIT WHEN c%NOTFOUND;
IF thePrice < 3.00 THEN
UPDATE Sells SET price = thePrice +
1.00;
WHERE bar = ’Joe’’s Bar’ AND beer =
theBeer; Note this is a SET clause
in an UPDATE, not an assignment.
END IF; PL/SQL uses := for assignments.
END LOOP; 27
CLOSE c;
 Tuple-Valued Variables
◆ PL/SQL allows a variable x to have
a tuple type.
◆ x R%ROWTYPE gives x the type of
R’s tuples.
◆ R could be either a relation or a
cursor.
◆ x.a gives the value of the
component for attribute a in the
tuple x. 28
 Example: Tuple Type
◆ Here are the declarations of JoeGouge(),
using a variable bp whose type is beer-
price pairs, as returned by cursor c.
CREATE OR REPLACE PROCEDURE
JoeGouge() AS
CURSOR c IS
SELECT beer, price FROM Sells
WHERE bar = ’Joe’’s Bar’;
bp c%ROWTYPE;
29
JoeGouge() Body Using bp
BEGIN
OPEN c;
LOOP
FETCH c INTO bp;
EXIT WHEN c%NOTFOUND;
IF bp.price < 3.00 THEN
UPDATE Sells SET price = bp.price + 1.00
WHERE bar = ’Joe’’s Bar’ AND beer
=bp.beer;
END IF; Components of bp are
END LOOP; obtained with a dot and
the attribute name
CLOSE c; 30

END;