15.3.4.1 Nested block terminology

A PL/SQL block nested within another PL/SQL block may be called by any of the following: nested block, enclosed block, child block or sub-block.

A PL/SQL block that calls another PL/SQL block (anonymous or named) may be referred to as either the enclosing block or the parent block.

15.3.4.2 Nested blocks provide scope

The general advantage of a nested block is that you create a scope for all the declared objects and executable statements in that block. You can use this scope to improve your control over activity in your program. For a discussion, see Section 15.3.5, "Scope and Visibility" later in this chapter.

Consider the following procedure, in which the president and vice-president of the specified company request their salaries be cut in half if their current salaries are more than ten times the average salary in the company. It's the sort of incentive plan that would encourage executives to "share the wealth."

PROCEDURE update_management
   (company_id_in IN NUMBER, avgsal_in IN NUMBER)
IS
BEGIN
   -- The vice-president shows his generosity...
   BEGIN
      SELECT salary INTO v_sal
        FROM employee
       WHERE company_id = company_id_in
         AND title = 'VICE-PRESIDENT';
      IF v_sal > avgsal_in * 10
      THEN
         UPDATE employee SET salary := salary * .50
          WHERE company_id = company_id_in
            AND title = 'VICE-PRESIDENT';
      ELSE
         DBMS_OUTPUT.PUT_LINE ('The VP is OK!');
      END IF;
   EXCEPTION
      WHEN NO_DATA_FOUND THEN NULL;
   END;

   -- The president shows her generosity...
   BEGIN
      SELECT salary INTO v_sal
        FROM employee
       WHERE company_id = company_id_in
         AND title = 'PRESIDENT';
      IF v_sal > avgsal_in * 10
      THEN
         UPDATE employee SET salary := salary * .50
          WHERE company_id = company_id_in
            AND title = `PRESIDENT';
      ELSE
         DBMS_OUTPUT.PUT_LINE ('The Prez is a pal!');
      END IF;
   EXCEPTION
      WHEN NO_DATA_FOUND THEN NULL;
   END;
END;

Each of the two SELECT-UPDATE combinations is embedded in its own anonymous PL/SQL block. Each block has its own exception section. Why go to all this trouble? Why couldn't the programmer just create a little script that will update the salary for a specified title? The following statement, saved to the updemp.sql file, would "do the trick" (&N is the syntax used to supply arguments to a SQL*Plus script):

UPDATE employee SET salary := salary * .50
 WHERE company_id = &1
   AND title = '&2'
   AND salary > &3 * 10;

and then execute the SQL script in SQL*Plus using the START command:

SQL> start updemp 1100 VICE-PRESIDENT
SQL> start updemp 1100 PRESIDENT

The programmer who was assigned this task took several things into account:

• The executives might decide they will want to take such actions repeatedly in the coming years; better to package the steps into a reusable chunk of code like a procedure than simply execute a series of SELECT-UPDATEs in SQL*Plus.

• Executives come and go frequently at the company. There is no guarantee that there will be a person in the employee table with a title of `PRESIDENT' or `VICE-PRESIDENT' at any given time. The first assumption argues for the encapsulation of these two SELECT-UPDATE steps into a procedure. This second assumption results in the need for embedded PL/SQL blocks around each UPDATE statement. Suppose that the procedure update_management did not make use of the embedded blocks. The code would then look like this:

  PROCEDURE update_management
     (company_id_in IN NUMBER, avgsal_in IN NUMBER)
  IS
  BEGIN
     SELECT salary INTO v_sal
       FROM employee
      WHERE company_id = company_id_in
        AND title = 'VICE-PRESIDENT';
     IF v_sal > avgsal_in * 10
     THEN
        UPDATE employee SET salary := salary * .50
         WHERE company_id = company_id_in
           AND title = 'VICE-PRESIDENT';
     ELSE
        DBMS_OUTPUT.PUT_LINE ('The VP is OK!');
     END IF;
     SELECT salary INTO v_sal
       FROM employee
      WHERE company_id = company_id_in
        AND title = 'PRESIDENT';
     IF v_sal > avgsal_in * 10
     THEN
        UPDATE employee SET salary := salary * .50
         WHERE company_id = company_id_in
           AND title = 'PRESIDENT';
     ELSE
        DBMS_OUTPUT.PUT_LINE ('The Prez is a pal!');
     END IF;
  END;

If there is a record in the employee table with the title of "VICE-PRESIDENT" (for the appropriate company_id) and if there is a record in the employee table with the title of "PRESIDENT," then this procedure works just fine. But what if there is no record with a title of "VICE-PRESIDENT"? What if the Vice-President did not want to take a 50% cut in pay and instead quit in disgust?

When the WHERE clause of an implict SELECT statement does not identify any records, PL/SQL raises the NO_DATA_FOUND exception (for more details on this phenomenon see Chapter 6, Database Interaction and Cursors, and pay particular attention to the sections concerning implicit cursors). There is no exception handler section in this procedure. As a result, when there is no Vice-President in sight, the procedure will immediately complete (well, abort, actually) and transfer control to the calling block's exception section to see if the NO_DATA_FOUND exception is handled there. The SELECT-UPDATE for the President's salary is therefore not executed at all.

NOTE: You don't actually need a BEGIN-END block around the second SELECT-UPDATE. This is the last statement, and if it fails nothing will be skipped. It is good practice, however, to include an exception statement for the procedure as a whole to trap errors and handle them gracefully.

Our programmer would hate to see an executive's wish go unfulfilled, so we need a mechanism to trap the failure of the first SELECT and allow PL/SQL procedure execution to continue on to the next SELECT-UPDATE. The embedded anonymous block offers this protection. By placing the first SELECT within a BEGIN-END envelope, you can also define an exception section just for the SELECT. Then, when the SELECT fails, control is passed to the exception handler for that specific block, not to the exception handler for the procedure as a whole (see Figure 15.7). If you then code an exception to handle NO_DATA_FOUND which allows for continued processing, the next SELECT-UPDATE will be executed.

Figure 15.7: An anonymous block "catches" the exception

15.3.4.3 Named modules offer scoping effect of nested block

You can also restructure the update_management procedure to avoid the problems addressed in the previous section, and also the redundancy in the code for that procedure. The following version of update_management uses an explicit cursor to avoid the problem of the NO_DATA_FOUND exception being raised. It also encapsulates the repetitive logic into a local or nested procedure.

PROCEDURE update_management
   (company_id_in IN NUMBER, avgsal_in IN NUMBER, decr_in IN NUMBER)
IS
   CURSOR salcur (title_in IN VARCHAR2)
   IS
      SELECT salary
        FROM employee
       WHERE company_id = company_id_in
         AND title = title_in
         AND salary > avgsal_in * 10;

   PROCEDURE update_exec (title_in IN VARCHAR2)
   IS
      salrec salcur%ROWTYPE;
   BEGIN
      OPEN salcur (title_in);
      FETCH salcur INTO salrec;
      IF salcur%NOTFOUND
      THEN
         DBMS_OUTPUT.PUT_LINE ('The ' || title_in || ' is OK!');
      ELSE
         UPDATE employee SET salary := salary * decr_in
          WHERE company_id = company_id_in
            AND title = title_in;
      END IF;
      CLOSE salcur;
   END;
BEGIN
   update_exec ('VICE-PRESIDENT');
   update_exec ('PRESIDENT');
END;

This final version also offers the advantage of consolidating the two SELECTs into a single explicit cursor, and the two UPDATEs into a single statement, thereby reducing the amount of code one would have to test and maintain. Whether you go with named or anonymous blocks, the basic concept of program (and therefore exception) scope remains the same. Use the scoping and visibility rules to your advantage by isolating areas of code and cleaning up the program flow.

15.3.5.1 Qualified identifiers

A qualifier for an identifier can be a package name, module name (procedure or function), or loop label. You qualify the name of an identifer with dot notation, the same way you would qualify a column name with the name of its table.

The names of the following identifiers are qualified:

:GLOBAL.company_id

A global variable in Oracle Forms

std_types.dollar_amount

A subtype declared in a package

The scope of an identifier is generally the block in which that identifier is declared. The following anonymous block declares and then references two local variables:

DECLARE
   first_day DATE;
   last_day DATE;
BEGIN
   first_day := SYSDATE;
   last_day := ADD_MONTHS (first_day, 6);
END;

Both the first_day and last_day variables are visible in this block. When I reference the variables in the assignment statements, I do not need to qualify their names. Also, the scope of the two variables is precisely this anonymous block.

I cannot make reference to either of those variables in a second anonymous block or in a procedure. Any attempt to do so will result in a compile failure, because the reference to those variables cannot be resolved.

If an identifier is declared or defined outside of the current PL/SQL block, it is visible (can be referenced without a qualifier) only under the following conditions:

• The identifier is the name of a standalone procedure or function on which you have EXECUTE privilege. You can then include a call to this module in your block.

• The identifier is declared in a block which encloses the current block.

15.3.5.2 Scope and nested blocks

Let's take a closer look at nested blocks and the impact on scope and visibility. When you declare a variable in a PL/SQL block, then that variable is local to that block, but is visible in or global to any blocks defined within the first, enclosing block. For example, in the following anonymous block, the variable last_date can be referenced anywhere inside the block:

/* The enclosing or outer anonymous block. */
DECLARE
   last_date DATE;
BEGIN
   last_date := LAST_DAY (SYSDATE);

   /* The inner anonymous block. */
   BEGIN
      IF last_date > :employee.hire_date
      THEN
         ...
      END IF;
   END;
   ...
END;

Even though last_date is not defined in the inner block, it can still be referenced there without qualification because the inner block is defined inside the outer block. The last_date variable is therefore considered a global variable in the inner block.

Figure 15.8 shows additional examples illustrating the concept of scope in PL/SQL blocks.

Figure 15.8: Scope of identifiers in PL/SQL blocks

15.3.5.3 Qualifying identifier names with module names

When necessary, PL/SQL offers many ways to qualify an identifier so that a reference to the identifier can be resolved. Suppose I create a package called company_pkg and declare a variable named last_company_id in that package's specification, as follows:

PACKAGE company_pkg
IS
   last_company_id NUMBER;
   ...
END company_pkg;

Then, when I reference that variable outside of the package, I must preface the identifer name with the package name:

IF new_company_id = company_pkg.last_company_id THEN ...

Because a variable declared in a package specification is global in your session, the last_company_id variable can be referenced in any program, but it is not visible unless it is qualified.

I can also qualify the name of an identifier with the module in which it is defined:

PROCEDURE calc_totals
IS
   salary NUMBER;
BEGIN
   ...
   DECLARE
      salary NUMBER;
   BEGIN
      salary := calc_totals.salary;
   END;
   ...
END;

The first declaration of salary creates an identifier whose scope is the entire procedure. Inside the inner anonymous block, however, I declare another identifer with the same name. So when I reference the variable "salary" inside the inner block, it will always be resolved first against the declaration in the inner block, where that variable is visible without any qualification. If I wish to make reference to the procedure-wide salary variable inside the inner block, I must qualify that variable name with the name of the procedure.

PL/SQL goes to a lot of trouble and has established many rules for determining how to resolve such naming conflicts. While it is good to be aware of such issues, you would be much better off never having to rely on these guidelines. Use unique names for your identifiers in different blocks so that you can avoid naming conflicts altogether.

15.3.5.4 Cursor scope

To use a cursor, you must declare it. But you cannot refer to that cursor -- whether to OPEN, CLOSE, or FETCH from it -- unless that cursor is accessible in your current PL/SQL block. The scope of a cursor is that part of a PL/SQL program in which you can refer to the cursor.

You can refer to a cursor in the code block in which it was declared and in all blocks defined within that declaring block. You cannot refer to the cursor outside of the declaring block unless the cursor is declared in a package (see Chapter 16, Packages, for more information on global variables and cursors). For example, if a cursor is declared in the get_employees procedure (see below) and within get_employees a second PL/SQL block is defined to calculate total compensation for each employee, the cursor and its attributes may still be referenced in that calculation block:

PROCEDURE get_employees
IS
   CURSOR emp_cur IS
      SELECT employee_id, sal + bonus FROM employee;
BEGIN
   OPEN emp_cur;
   DECLARE
      empid NUMBER;
      total_comp NUMBER;
   BEGIN
      FETCH emp_cur INTO empid, total_comp;
      IF total_comp < 5000
      THEN
         MESSAGE (' I need a raise!');
      END IF;
   END;
END;

If, on the other hand, the cursor is declared within an inner block, then that cursor cannot be referenced in an outer block. In the next procedure, the outer block declares the variables that receive the data fetched from the cursor. The inner block declares the cursor. While the inner block can open the cursor without error, the FETCH statement is outside the scope of the cursor-declaring block and will therefore fail:

PROCEDURE get_employees IS
   empid NUMBER;
   total_comp NUMBER;
BEGIN
   DECLARE
      CURSOR emp_cur IS
         SELECT employee_id, sal + bonus FROM employee;
   BEGIN
      OPEN emp_cur;
   END;

   /* This fetch will not be able to identify the cursor. */

   FETCH emp_cur INTO empid, total_comp; -- INVALID!
   IF total_comp < 5000
   THEN
      MESSAGE (' I need a raise!');
   END IF;
END;

The rules for cursor scope are, therefore, the same as those for all other identifiers.