my $file = "c:/webcod_enf.txt";
open F, $file or die $!;

local $/ = "\n\n";  # break up into rough 'records' at blank lines
                    # often you can get good records just by setting $/
                    # not with this data though ;-)

while (<F>) {

    # skip the extraneous data, valid chunks will start with ^\s*\d{5}

    unless ( m/^\s*\d{5}/) {
        #print "Skipping:\n$_";
      next;
    }

    # now we have chunks of real data to parse. we split it on the unique m/^\s*\d{5}/m
    # numeric feature to break out the individual records. we use a lookahead assertion
    # to do this so we don't loose that data in the split

    for my $course( split /(?=^\s*\d{5})/m, $_ ) {
      next if $course =~ m/^\s*$/;   # we possibly get a null record to start so skip
        print "$course\n\n";
    }
}

__DATA__
 92861 APMA 109  0001 GI LC CALCULUS I                        4.0
       0900-0950 M W F  OLS   011  OBERHAUSER      JP 055 002 O
       0830-0920  T     OLS   005


 90063 APMA 109  0002 GI LC CALCULUS I                        4.0
       1000-1050 M W F  OLS   120  BECK            M  055 004 O
       0830-0920    R   OLS   120


 91589 APMA 109  0003 GI LC CALCULUS I                        4.0
       1100-1150 M W F  OLS   120  BECK            M  055 006 O
       0830-0920  T     MEC   205