On Tue, Aug 19, 2003 at 05:54:42PM -0500, Dan Muey wrote:
> Howdy all:
>
> I'm trying to figure out the best way to test a string agains a list of regexs like
> so:
>
> my @regex = qw(qr(joe$) qr(^mama) qr([abc]));
As was pointed out already, don't use the qw().
Here are some interesting benchmarks. The 'docs_in_col' file was
about 16k, and the strings I was testing for were right at the bottom.
#!/usr/bin/perl
use warnings;
use strict;
use IO::File;
use Benchmark;
my $fh = new IO::File("docs_in_col") or die $!;
my $str;
{
local $/;
$str = <$fh>;
}
my $alt_re = qr(Zucker|Zuckerman|Zurrow);
my @grep_re = (qr(Zucker), qr(Zuckerman), qr(Zurrow));
timethese(5000,
{
match_alts_scalar => \&match_alts_scalar,
match_alts_array => \&match_alts_array,
grep_mults => \&grep_mults,
}
);
###----------------------------------------------------
sub match_alts_scalar {
my $found = ($str =~ /$alt_re/);
return $found;
}
###----------------------------------------------------
sub grep_mults {
my $found = grep { $str =~ /$_/ } @grep_re;
return $found;
}
And here are the results:
Benchmark: timing 5000 iterations of grep_mults, match_alts_scalar...
grep_mults: 1 wallclock secs ( 0.54 usr + 0.00 sys = 0.54 CPU) @ 9275.36/s (n=5000)
match_alts_scalar: 40 wallclock secs (39.20 usr + 0.02 sys = 39.22 CPU) @ 127.49/s
(n=5000)
This should only be considered a first approximation, since there are
various things I'm not controlling for (all my strings were at the
end, they were fixed strings, etc).
Still, I would use the grep version.
--Dks
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
