SQL 101: Ultra-Flexible Cursors—The Beauty of PREPARE

SQL
Typography
  • Smaller Small Medium Big Bigger
  • Default Helvetica Segoe Georgia Times

SQL’s PREPARE statement, combined with ILE RPG, can bring unmatched flexibility to your application’s code. Read on to find how you can do it. It’s easier than you think!

Using thePREPARE SQL instruction, you can have a single cursor and use it according to conditions defined by user input, for instance. I’ve seen subfiles with different columns, selections, and sorting being managed with a single cursor. How, you might ask? The PREPARESQL instruction allows you to define a cursor loosely:

/Free

   Exec SQL

     DECLARE mainCursor CURSOR

     FOR MY_SELECT;

MY_SELECT is, or rather, will be the SELECT statement that the cursor will use, but at declaration time, MY_SELECT is only a placeholder. You need to define its “variable” content with a PREPARE statement before the DECLARE statement, but you’re free to compose it in any way that you want, as long as you specify a valid SELECT statement. Here’s how you do it:

/Free

   W_My_SQL_Stmt = ‘SELECT * FROM InvMst WHERE WHID = ‘ +

                   %CHAR(K_WHID) + ‘ AND ExpDate < CurDate’;

   Exec SQL

     PREPARE MY_SELECT FROM :W_MySQL_Stmt;

These two statements will assign a dynamic statement to the W_My_SQL_Stmt variable. Note that it can be even more dynamic; I could change the columns, the sort, or any other part of the SELECT statement. The resulting string is then transformed into an SQL “thing” that I’ll use in the DECLARE statement to define my cursor. After this operation, the usual OPEN, FETCH, CLOSE cursor procedure is used.

Here’s a practical example: Imagine you want a function that returns the total number of records of a given table or table join. The function is abstract, in the sense that the statement describing the table or table join is not known beforehand and neither are the conditions (the WHERE part of the statement) in which it’ll be run. This is the perfect example of PREPARE’s flexibility. As usual, let’s start with the function’s prototype interface:

*-------------------------------------------------------------------*

* Procedure: RowCounter - return a row count from a select statement*

*-------------------------------------------------------------------*

PRowCounter       B                   EXPORT

DRowCounter       PI                 Like(W_Out)

* Input Parms

DP_Stm                       1000A   Value

* Work variables

DW_Out           S             12 0

Yes, I know, it should be a /FREE definition. It’s not because I’m reusing an actual piece of production code that hasn’t been converted to free-format.

Anyway, there’s nothing really new here; just notice that the input parameter is a regular string. This string will hold a complete (and preferably valid) SELECT statement with an aggregation function in the SELECT clause. Although this is designed to use a SELECT COUNT(*) statement, you can use it with other aggregation functions as well, such as SUM, for instance.

Note that the string’s size (1,000 characters) should be enough for most SQL statements, but if you plan to use complex (or simply very long) SELECT statements, it might not be enough. If you get strange “invalid SQL statement” errors at runtime, you should look into this variable’s size and compare it with the value you’re trying to pass it. There’s only one work variable, W_Out, which will be used to return the output of the SELECT statement, as you’ll see in a minute:

/FREE

     EXEC SQL

       PREPARE S1 FROM :P_Stm;

     EXEC SQL

       DECLARE C1 CURSOR FOR S1;

     EXEC SQL

       OPEN C1;

     EXEC SQL

       FETCH C1 INTO :W_Out;

     EXEC SQL

       CLOSE C1;

This is the main body of the function. This is an RPG function, but there’s only SQL in it. Let’s analyze the code. The input parameter, P_Stm, is used to create an “SQL thing” named S1 that is going to be used by the C1 cursor’s DECLARE statement as its SELECT part. From here, the code follows the usual cursor operation sequence: OPEN, FETCH, and CLOSE. All that is left to do is return W_Out to the outside world and end the function:

     Return W_Out;

/END-FREE

PRowCounter       E

Note that the function doesn’t know (and doesn’t care) what the SELECT statement looks like, as long as it’s a valid SELECT statement that returns a number—the typical result of an aggregation function. PREPARE allows you to create this type of “hollow function” with dynamic SELECT statements, but it can also be used for other dynamic statements, as you’ll see later.

This is a simple example, but I believe it gets the point across. With the PREPARE statement, you can write generic RPG functions that can be reused in several different situations within the same type of scenario (for instance, quickly returning the maximum value that fulfills certain very specific conditions for a variety of different files/variations within the same file) or even fully dynamic Open Query File replacements, as mentioned in the previous TechTip.

That’s all the time we have this time around. In the next article, I’ll be discussing how to embed other SQL statements in RPG…and why you should do it.

BLOG COMMENTS POWERED BY DISQUS