Find IF as first word

[Michael Simpson]

I have a mainframe script that checks the level of nesting in cobol programs (so people get a slap on the wrist for coding something like if else if else if else if else if else etc etc)

/* Remove any commented lines from the FIND command */    
'exclude "*" .strts .ends all 'leading leading            
                                                          
/* Count only NON-commented IF's                    */    
'find " IF " .strts .ends all nx'                         
"(res) = find_counts"                                     

This has worked fine, until it hit the following code

string                                              
  '*** You have defined default values for '        
                           delimited by size        
  '(character) field '                              
                           delimited by size        
  def-col-name(ix-1)       delimited by ' '         
  ' with BN (as if it were numeric). Use = instead' 
                           delimited by size        
  into shbgg015-meddelande-tab                      
end-string                                          

whereupon it finds the IF in quotes and treats it as if it was part of an IF/END-IF statement.

I could easily change the

'find " IF " .strts .ends all nx'    

to perform a loop, but is there a variation on the find command that would do the job for me (basically, I only want to count the word IF if it's the first word on the line)

[dbzthedinosaur]

part of the FIND command syntax is FIRST and WORD

[Nic Clouston]

But FIRST refers to the first OCCURRENCE on a line, not the first word so the 'IF' could be the last word on the line and still be in the result set. WORD just makes sure that only the WORD 'IF' is included in the result set and not words such as 'dIFficult'.

You may have to check the POS of any ' and see if the IF comes after it and before a second '.

Or, find the position of the first non-blank character and see if it is IF.

[Michael Simpson]

that's about what I'd figured - have to perform a loop.

Just wanted to check that there wasn't some finesse I didn't know about

[dbzthedinosaur]

I stand corrected, FIND parms do not fit as I suggested.

I don't use FIND that often basically due to this problem. have to check the columns, etc...

normally for this kind of task, I use a combination of LINE to populate a stem, and PARSE to detemine if the 'IF' is really an IF.

[dbzthedinosaur]

enforcing standards is always alot of fun.

when the site could not afford (or would not buy) standards enforcing software
(it is expensive)
I would write a COBOL program that would insert a non astriked '*' comment in the source explaining the offense.
This program was part of the compile proc, that everyone had to use,
and of course the un-astriked comment(s) would cause a compile failure.

this allowed for consequences - can't compile the program
and avoided such issues as:

• a script has to be invoked either by the user or you
• 'the program works and needs to go to production, now'

[RonB]

Originally Posted by Michael Simpson

(basically, I only want to count the word IF if it's the first word on the line)

What would you want to do if legitimate 'IF'(s) are not the first word(s) on (a) line(s), for example

    IF VAR-X > VAR-Y
       THEN IF VAR-Y = VAR-Z
               THEN PERFORM PARA-A
               ELSE IF VAR-Y > VAR-Z
                       THEN PERFORM PARA-B
                       ELSE PERFORM PARA-C
                    END-IF
            END-IF
       ELSE PERFORM PARA-D
    END-IF

[don.leahy]

Originally Posted by RonB

What would you want to do if legitimate 'IF'(s) are not the first word(s) on (a) line(s), for example

    IF VAR-X > VAR-Y
       THEN IF VAR-Y = VAR-Z
               THEN PERFORM PARA-A
               ELSE IF VAR-Y > VAR-Z
                       THEN PERFORM PARA-B
                       ELSE PERFORM PARA-C
                    END-IF
            END-IF
       ELSE PERFORM PARA-D
    END-IF

The above is a good example of why the commercial software products that enforce standards are so expensive. To do this rigourously requires you to re-engineer the Cobol language parser used by the compiler and then apply your local standards. A challenging undertaking, so say the least.

[dbzthedinosaur]

I would bounce the code because it violated standards.
I work with a lot of doctorates (physics)
and they also have no sensitivity to readability.

THEN, ELSE also alone on a separate line

    IF VAR-X > VAR-Y
    THEN 
       IF VAR-Y = VAR-Z
       THEN 
           PERFORM PARA-A
       ELSE 
           IF VAR-Y > VAR-Z
           THEN 
               PERFORM PARA-B
           ELSE 
               PERFORM PARA-C
           END-IF
       END-IF
    ELSE 
       PERFORM PARA-D
    END-IF

[RonB]

FWIW, I, myself, would NEVER write COBOL code in the manner that I posted as an "example" of the kind of code that "might" be encountered in someone else's source code. I merely posted it as an example of another kind of code that Michael "might" encounter that may cause his filter to hiccup or misfire.

In his preamble he, himself, used the phrase "coding something like if else if else if else if else if else etc etc" - a phrase in which there are multiple IF's following the word ELSE on the same line.