Re: Improving dismal performance - Part 1

[rir]

This isn't complete, but shows a common approach.

Be well
rir

# XXX lots of possibly magic string literals

$| = 1;
local $_;

while (<DATA>) {
    # find next record
    next unless /^\s*Record /;

    # read a record

    $_ = '';
    my @record;
    do {
        s/^ *//;
        push @record, $_ if $_  && $_ !~ /^\s*End of Sub/ && $_ !~ /^\
+s*Sub/;
        $_ = <DATA>;
    }while ( $_ !~ /^\s*End of Record/ );


    # massage the raw data in @record
    # XXX incomplete

    for my $item ( @record ) {
        $item =~ s/" = "/ /;     
        $item =~ s/"//g;
        $item = "F " . $item;
    }
    @record = sort @record;


    # output preamble
    print STDOUT "RECORD\n";
    print STDOUT "#addkey\n";
    print STDOUT "#filename FF\n";
    # XXX more?

    # output @record
    print STDOUT @record;

    # output trailer
    print STDOUT ".\n";
}

[PoorLuzer]

Yes, the script before this one (using Tie) looked exactly like this, and that is still what I use now after the new version failed to do anything useful in reasonable time, but I was hoping that maybe I could optimize this code somehow - using Tie makes things so easy - just like looping through an array!

Well, I am not totally out of luck, as I am reading line by line anyways, and not much change would be required to translate it into a direct file IO .. but in case someone comes across a light bulb in their vicinity, keep this node updated!