Exception Handling in PL/SQL

Last Updated : 28 Apr, 2018

An exception is an error which disrupts the normal flow of program instructions. PL/SQL provides us the exception block which raises the exception thus helping the programmer to find out the fault and resolve it.

There are two types of exceptions defined in PL/SQL

User defined exception.
System defined exceptions.

Syntax to write an exception

WHEN exception THEN 
    statement;

DECLARE
declarations section;

BEGIN
executable command(s);

EXCEPTION
WHEN exception1 THEN
statement1;
WHEN exception2 THEN
statement2;
[WHEN others THEN]
/* default exception handling code */

END;

Note:
When other keyword should be used only at the end of the exception handling block as no exception handling part present later will get executed as the control will exit from the block after executing the WHEN OTHERS.

System defined exceptions:
These exceptions are predefined in PL/SQL which get raised WHEN certain database rule is violated.
System-defined exceptions are further divided into two categories:

Named system exceptions.
Unnamed system exceptions.

Named system exceptions: They have a predefined name by the system like ACCESS_INTO_NULL, DUP_VAL_ON_INDEX, LOGIN_DENIED etc. the list is quite big.

So we will discuss some of the most commonly used exceptions:

Lets create a table geeks.

create table geeks(g_id int , g_name varchar(20), marks int); 
insert into geeks values(1, 'Suraj',100);
insert into geeks values(2, 'Praveen',97);
insert into geeks values(3, 'Jessie', 99);

NO_DATA_FOUND: It is raised WHEN a SELECT INTO statement returns no rows. For eg:

DECLARE
   temp varchar(20); 
  
BEGIN
   SELECT g_id into temp from geeks where g_name='GeeksforGeeks'; 
  
exception 
   WHEN no_data_found THEN
      dbms_output.put_line('ERROR'); 
      dbms_output.put_line('there is no name as'); 
      dbms_output.put_line('GeeksforGeeks in geeks table'); 
end; 

Output:

ERROR
there is no name as GeeksforGeeks in geeks table

TOO_MANY_ROWS:It is raised WHEN a SELECT INTO statement returns more than one row.

DECLARE
   temp varchar(20); 
  
BEGIN
  
-- raises an exception as SELECT  
-- into trying to return too many rows 
   SELECT g_name into temp from geeks; 
   dbms_output.put_line(temp); 
  
EXCEPTION 
   WHEN too_many_rows THEN
      dbms_output.put_line('error trying to SELECT too many rows'); 
  
end; 

Output:

error trying to SELECT too many rows

VALUE_ERROR:This error is raised WHEN a statement is executed that resulted in an arithmetic, numeric, string, conversion, or constraint error. This error mainly results from programmer error or invalid data input.

DECLARE
   temp number;    
  
BEGIN
   SELECT g_name  into temp from geeks where g_name='Suraj'; 
   dbms_output.put_line('the g_name is '||temp); 
  
EXCEPTION 
   WHEN value_error THEN
   dbms_output.put_line('Error'); 
   dbms_output.put_line('Change data type of temp to varchar(20)'); 
  
END; 

Output:

Error
Change data type of temp to varchar(20)

ZERO_DIVIDE = raises exception WHEN dividing with zero.

DECLARE
   a int:=10; 
   b int:=0; 
   answer int; 
  
BEGIN
   answer:=a/b; 
   dbms_output.put_line('the result after division is'||answer); 
  
exception 
   WHEN zero_divide THEN
      dbms_output.put_line('dividing by zero please check the values again'); 
      dbms_output.put_line('the value of a is '||a); 
      dbms_output.put_line('the value of b is '||b); 
END; 

Output:

dividing by zero please check the values again
the value of a is 10
the value of b is 0

Unnamed system exceptions:Oracle doesn’t provide name for some system exceptions called unnamed system exceptions.These exceptions don’t occur frequently.These exceptions have two parts code and an associated message.
The way to handle to these exceptions is to assign name to them using Pragma EXCEPTION_INIT
Syntax:

PRAGMA EXCEPTION_INIT(exception_name, -error_number);

error_number are pre-defined and have negative integer range from -20000 to -20999.

Example:

DECLARE
   exp exception; 
   pragma exception_init (exp, -20015); 
   n int:=10; 
  
BEGIN 
   FOR i IN 1..n LOOP 
      dbms_output.put_line(i*i); 
         IF i*i=36 THEN
            RAISE exp; 
         END IF; 
   END LOOP; 
  
EXCEPTION 
   WHEN exp THEN
      dbms_output.put_line('Welcome to GeeksforGeeks'); 
  
END; 

Output:

1
4
9
16
25
36
Welcome to GeeksforGeeks

User defined exceptions:
This type of users can create their own exceptions according to the need and to raise these exceptions explicitly raise command is used.

Example:

Divide non-negative integer x by y such that the result is greater than or equal to 1.

From the given question we can conclude that there exist two exceptions

Division be zero.
If result is greater than or equal to 1 means y is less than or equal to x.

DECLARE
   x int:=&x; /*taking value at run time*/ 
   y int:=&y; 
   div_r float; 
   exp1 EXCEPTION; 
   exp2 EXCEPTION; 
  
BEGIN
   IF y=0 then
       raise exp1; 
  
   ELSEIF y > x then
      raise exp2; 
  
   ELSE
      div_r:= x / y; 
      dbms_output.put_line('the result is '||div_r); 
  
   END IF; 
  
EXCEPTION 
   WHEN exp1 THEN
      dbms_output.put_line('Error'); 
      dbms_output.put_line('division by zero not allowed'); 
  
   WHEN exp2 THEN
      dbms_output.put_line('Error'); 
      dbms_output.put_line('y is greater than x please check the input'); 
  
END; 

Input 1: x = 20
         y = 10

Output: the result is 2

Input 2: x = 20
         y = 0

Output:
Error
division by zero not allowed

Input 3: x=20
         y = 30

Output:<.em>
Error
y is greater than x please check the input

RAISE_APPLICATION_ERROR:
It is used to display user-defined error messages with error number whose range is in between -20000 and -20999. When RAISE_APPLICATION_ERROR executes it returns error message and error code which looks same as Oracle built-in error.

Example:

DECLARE

    myex EXCEPTION;

    n NUMBER :=10;



BEGIN

    FOR i IN 1..n LOOP

    dbms_output.put_line(i*i);

        IF i*i=36 THEN

        RAISE myex;

        END IF;

    END LOOP;



EXCEPTION

    WHEN myex THEN

        RAISE_APPLICATION_ERROR(-20015, 'Welcome to GeeksForGeeks');



END;

Output:

Error report: ORA-20015: Welcome to GeeksForGeeks ORA-06512: at line 13 1 4 9 16 25 36

Note: The output is based on Oracle Sql developer, the output order might change IF you’re running this code somewhere else.

Scope rules in exception handling:

We can’t DECLARE an exception twice but we can DECLARE the same exception in two dIFferent blocks.

Exceptions DECLAREd inside a block are local to that block and global to all its sub-blocks.

As a block can reference only local or global exceptions, enclosing blocks cannot reference exceptions DECLAREd in a sub-block.
If we reDECLARE a global exception in a sub-block, the local declaration prevails i.e. the scope of local is more.

Example:

DECLARE

   GeeksforGeeks EXCEPTION;

   age NUMBER:=16;

BEGIN



   -- sub-block BEGINs

   DECLARE



      -- this declaration prevails

      GeeksforGeeks EXCEPTION;

      age NUMBER:=22;



   BEGIN

      IF age > 16 THEN

         RAISE GeeksforGeeks; /* this is not handled*/

      END IF;



   END;

   -- sub-block ends



EXCEPTION

  -- Does not handle raised exception

  WHEN GeeksforGeeks THEN

    DBMS_OUTPUT.PUT_LINE

      ('Handling GeeksforGeeks exception.');



  WHEN OTHERS THEN

    DBMS_OUTPUT.PUT_LINE

      ('Could not recognize exception GeeksforGeeks in this scope.');

END;

Output:

Could not recognize exception GeeksforGeeks in this scope.

Advantages:

Exception handling is very useful for error handling, without it we have to issue the command at every point to check for execution errors:
Example:

Select .. .. check for 'no data found' error Select .. .. check for 'no data found' error Select .. .. check for 'no data found' error

Here we can see that it is not robust as error processing is not separated from normal processing and IF we miss some line in the code than it may lead to some other kind of error.

With exception handling we handle errors without writing statements multiple times and we can even handle dIFferent types of errors in one exception block:
Example:

BEGIN SELECT ... SELECT ... SELECT ... . . . exception WHEN NO_DATA_FOUND THEN /* catches all 'no data found' errors */ ... WHEN ZERO_DIVIDE THEN /* different types of */ WHEN value_error THEN /* errors handled in same block */ ...

From above code we can conclude that exception handling

Improves readability by letting us isolate error-handling routines and thus providing robustness.

Provides reliability, instead of checking for dIFferent types of errors at every point we can simply write them in exception block and IF error exists exception will be raised thus helping the programmer to find out the type of error and eventually resolve it.

Uses: One of the real lIFe use of exception can be found in online train reservation system.
While filling the station code to book the ticket IF we input wrong code it shows us the exception that the code doesn’t exist in database.

Reference: You can find the list of all pre-defined exception here.
Total number of pre-defined exceptions

Next Article
PostgreSQL - Assert

S

SrjSunny

Improve

Article Tags :

Technical Scripter