How can I speed up my Perl regex matching?

Question

I want to capture several text using the following regex:

$text_normal = qr{^(\/F\d+) FF (.*?) SCF SF (.*?) MV (\(.*?)SH$};

A sample of the string is like below:

my $text = '/F12345 FF FF this is SCF SF really MV (important stuff SH';

Can that be rewritten to speed up the matching?

Answer 1

There's no single answer to optimizing a regex. You can watch what a particular regex is doing with the re pragma:

 use re 'debugcolor';

Once you see what it traverses the string, you see where it is having problems and adjust your regex from there. You'll learn a little about the regex engine as you do that.

You should also check out Mastering Regular Expressions, which tells you how regular expressions work and why some patterns are slower than others.

Answer 2

Without seeing some sample data it is hard to say.

Generally, it is a good idea to avoid using .*. Look for any possible sources of unneeded backtracking, and eliminate them.

You might be able to get away with a split with a slice if your needs are simple.

 my @vals = (split / /, $string)[0,2,5,7];

Answer 3

This very much depends on the profile of the data you are scanning.

You identify the piece of your regular expression which filters out the most input and do a separate simpler regular expression for that expression.

For instance, if only 5% of your input date contained the 'MV' string, you could filter for this first and only apply the full more complex regular expression if the simpler one is true.

So you would have:

if ( $text_normal =~ / MV / ) {
  $text_normal = qr{^(\/F\d+) FF (.*?) SCF SF (.*?) MV (\(.*?)SH$};
  if .......
  }
}