On Tuesday, June 12, 2001, at 03:20 , David J. Iberri wrote:
> I know there are some cleaner ways to do this, but they slip my
> mind for the
> moment... :-)
Mind you, I ran these tests in Perl 5.6.1 on debian-powerpc, but
the results should be representative.
#!/usr/bin/perl -w
# macperltest
#use strict;
use Benchmark;
my $initial_string = "first value: <xml1> sec value: <xml2>";
my %hash = (
"<xml1>" => 33,
"<xml2>" => 44,
);
my $re = '(' . join('|', keys %hash) . ')';
my $pre = qr/$re/;
my $code1 = sub
{
my $string = $initial_string;
};
my $code2 = sub
{
my $string = $initial_string;
while (my ($key, $value) = each %hash)
{
$string =~ s/$key/$value/i;
}
};
my $code3 = sub
{
my $string = $initial_string;
$string =~ s/$re/$hash{$1}/gi;
};
my $code4 = sub
{
my $string = $initial_string;
$string =~ s/$pre/$hash{$1}/gi;
};
timethese($ARGV[0], {
empty => $code1,
loop => $code2,
regexp => $code3,
precomp => $code4,
});
1;
__END__
craigc@scooter:~$ ./macperltest 100000
Benchmark: timing 100000 iterations of empty, loop, precomp, regexp...
empty: 2 wallclock secs ( 0.67 usr + 0.00 sys = 0.67
CPU) @ 149253.73/s (n=100000)
loop: 104 wallclock secs (52.15 usr + 0.01 sys = 52.16
CPU) @ 1917.18/s (n=100000)
precomp: 72 wallclock secs (36.20 usr + 0.00 sys = 36.20
CPU) @ 2762.43/s (n=100000)
regexp: 76 wallclock secs (38.08 usr + 0.01 sys = 38.09
CPU) @ 2625.36/s (n=100000)
Conclusion: Using one big regular expression is much faster than
an explicit loop. Copying the initial string adds an
insignificant amount of time to each iteration.
--
Craig S. Cottingham
[EMAIL PROTECTED]
PGP key available from:
<http://pgp.ai.mit.edu:11371/pks/lookup?op=get&search=0xA2FFBE41>
ID=0xA2FFBE41, fingerprint=6AA8 2E28 2404 8A95 B8FC 7EFC 136F
0CEF A2FF BE41
