Introducing The Cursor

You use a cursor to manage a query result that may contain more than one row when you want to make all the qualifying rows available to the program user. Cursors are used in sequential table processing and BULK table processing, as shown later in this chapter.

Like the cursor on a terminal screen, an ALLBASE/SQL cursor is a position indicator. It does not, however, point to a column. Rather, it points to one row in an active set. An active set is a query result obtained when a SELECT command associated with a cursor (defined in a DECLARE CURSOR command) is executed (using the OPEN CURSOR command).

Each cursor used in a program must be declared before it is used. You use the DECLARE CURSOR command to declare a cursor. The DECLARE CURSOR command names the cursor and associates it with a particular SELECT command:

   EXEC SQL DECLARE CURSOR1

             CURSOR FOR

             SELECT PARTNAME,

                    SALESPRICE

               FROM PURCHDB.PARTS

              WHERE PARTNUMBER BETWEEN :LOWVALUE AND :HIGHVALUE

                ORDER BY PARTNAME

   END-EXEC.

All cursor names within one program must be unique. You use a cursor name when you perform data manipulation operations using the cursor.

The SELECT command in the cursor declaration does not specify any output host variables. The SELECT command can, however, contain input host variables, as in the WHERE clause of the cursor declaration above.

Rows in the active set are returned to output host variables when the FETCH command is executed:

   EXEC SQL OPEN CURSOR1 END-EXEC.

   .

   .       The OPEN command allocates internal

   .       buffer space for the active set.

   .

   EXEC SQL [BULK] FETCH CURSOR1 INTO OutputHostVariables END-EXEC.



           The FETCH command delivers one row or (if the

           BULK option is used) multiple rows of the active

           set into output host variables.

If a serial scan will be used to retrieve the active set, ALLBASE/SQL locks the table(s) when the OPEN command is executed. If an index scan will be used, locks are placed when rows are fetched.

Both the OPEN and the FETCH commands position the cursor:

The row at which the cursor points at any one time is called the current row. When a row is a current row, you can delete it as follows:

   EXEC SQL DELETE FROM PURCHDB.PARTS

                  WHERE CURRENT OF CURSOR1

   END-EXEC.

When you delete the current row, the cursor remains between the row deleted and the next row in the active set until you execute the FETCH command again:

   EXEC SQL FETCH  CURSOR1

             INTO :PARTNAME :PARTNAMEIND,

                  :SALESPRICE :SALESPRICEIND

   END-EXEC.

When a row is a current row you can update it if the cursor declaration contains a FOR UPDATE OF clause naming the column(s) you want to change. The following cursor, for example, can be used to update the SALESPRICE column of the current row by using the WHERE CURRENT OF option in the UPDATE command:

   EXEC SQL DECLARE CURSOR2

             CURSOR FOR

             SELECT PARTNAME, SALESPRICE

               FROM PURCHDB.PARTS

              WHERE PARTNUMBER BETWEEN :LOWVALUE AND :HIGHVALUE

              FOR UPDATE OF SALESPRICE

   END-EXEC.

   .

   .     Because the DECLARE CURSOR command is not

   .     executed at run time, no status checking code

   .     needs to appear here.

   .

   EXEC SQL OPEN CURSOR2 END-EXEC.

   .

   .     The program fetches and displays one row at a time.

   .

   EXEC SQL FETCH  CURSOR2

             INTO :PARTNAME :PARTNAMEIND,

                  :SALESPRICE :SALESPRICEIND

   END-EXEC.

   .

   .     If the program user wants to change the SALESPRICE

   .     of the row displayed (the current row), the UPDATE

   .     command is executed.  The new SALESPRICE entered by

   .     the user is stored in an input host variable named

   .     NewSALESPRICE.

   .

   EXEC SQL UPDATE PURCHDB.PARTS

               SET SALESPRICE = :NEWSALESPRICE

             WHERE CURRENT OF CURSOR2

   END-EXEC.



         After the UPDATE command is executed, the updated

         row remains the current row until the FETCH command

         is executed again.

The restrictions that govern deletions and updates using a view also govern deletions and updates using a cursor. You cannot delete or update a row using a cursor if the cursor declaration contains any of the following:

After the last row in the active set has been fetched, the cursor is positioned after the last row fetched and the value in SQLCODE is equal to 100. Therefore to retrieve all rows in the active set, you execute the FETCH command until SQLCODE = 100. In the following example, a flag named DONE-FETCH is set to X after the last row in the active set has been fetched, and fetching stops:

   77   DONE-FETCH-FLAG          PIC X VALUE SPACE.

     88    NOT-DONE-FETCH        VALUE SPACE.

     88    DONE-FETCH            VALUE 'X'.

   .

   .

   .

   PROCEDURE DIVISION.

   .

   .

   .

       PERFORM FETCH-ROW THRU FETCH-ROW-EXIT

               UNTIL DONE-FETCH.

   .

   .

   .

   FETCH-ROW.

   .

   .

   .

       EXEC SQL FETCH  CURSOR3

                 INTO :PARTNUMBER,

                      :PARTNAME :PARTNAMEIND,

                      :SALESPRICE :SALESPRICEIND

       END-EXEC.

       IF SQLCODE = 0 THEN PERFORM DISPLAY-ROW

       ELSE

       IF SQLCODE = 100

          MOVE 'X' TO DONE-FETCH-FLAG

          DISPLAY "Row not found or no more rows"

          GO TO FETCH-ROW-EXIT

       ELSE

          PERFORM SQL-STATUS-CHECK.

   FETCH-ROW-EXIT.

When you are finished operating on an active set, you use the CLOSE command:

   EXEC SQL CLOSE CURSOR3 END-EXEC.

When you close a cursor, the active set becomes undefined and you cannot use the cursor again unless you issue an OPEN command to reopen it. The COMMIT WORK and ROLLBACK WORK commands also close any open cursors, automatically.

Figure 6-2 summarizes the effect of the cursor related commands on the position of the cursor and on the active set. All the commands shown, plus the DECLARE CURSOR command, must be included within one preprocessed unit (main program or subprogram).

Figure 6-2 Effect of SQL Commands on Cursor and Active Sets

Chapter 8 contains more detailed information about using cursors. See Chapter 11 for examples of using the KEEP CURSOR option of the OPEN command.