I was dreaming when I wrote this, forgive me if it goes astray...

All in all, it's a pretty clear piece of code as is. If I were a contractor assigned to it as legacy code, I would be very thankful and think nice thoughts about the coder that wrote it.

Couple of things you could do:

• You could put the regexps in a subroutine called 'clean_line', and have them return the line without all the "formatting". Lose a bit of efficiency, gain a bit of clarity and modularity. Pulling stuff out into subroutines frequently makes them clearer.
• You could combine the first two '\s' lines into a single 's/^\s*(.*?)\s*$/$1/;' expression. Don't know if that would slow things down; might.
• At the risk of kicking off a religious war, you might want to put your comments on the line above the code, instead of to the right. Easier to maintain.
• 'use strict'. I don't think it would change anything, just good to have in there.
• The superscript-three up there would look like line noise to me, and I'd probably be tempted to delete it. Replace it with a backslash code.

So, here's your first input loop, rewritten as I've suggested:

while (<REPORT>) {

    $_ = clean_line($_);

    if ($_ =~ /\*+/) {
       $_ =~ s/\*+//;    #remove asterisk only rows;
       $count++;         #the asterisk line only happens once per reco
+rd,
    }                       #so we can use it to count records.
    if (($_ =~ /Cosmetics/)||($_ =~ /Function/)) { #look for _X_Pass o
+r _X_Fail
       $_ =~ s/.+_X_(.{4}).*/$1/;   #make it Pass or Fail
    }
    print OUTFILE "$_";   #print the line to the intermediate file.
 }

##
## clean_line()
##
## Argument: $line (string) - Line from the file that needs cleaning
##
## Returns: string - cleaned-up line
##
## Clean_line() returns the $line minus leading and
## trailing whitespace, headers, etc.
##
sub clean_line
{
   my $line = shift;

    #remove the leading and trailing whitespace
    $line =~ s/^\s*(.*?)\s*$/$1/;             
    tr/\n\r³//;              #remove all carriage returns;
    s/.+ : //;           #remove the header information;
}
[download]

Something like that... my brain isn't running. Also, you could consider reading the input file record-at-a-time by setting the input record separator to '*' x (whatever number of stars they put in the input file) . "\n". Then you wouldn't need to watch for the star-lines and count them separately. You might need to modify your regexps to deal with that though.

stephen