split

[Christian Kruse]

Hoi,

Naja, eigentlich schon, aber 'patschert' ist es trotzdem:
($straße,$hausnummer) = split(/\s*([0-9]+)/,$wohnort,2);

[...]

($straße, $hausnummer) = split /(?=[0-9])/, $wohnort, 2;

Hui, das ist interessant. Der Algorithmus ist sehr schnell:

#!/usr/bin/perl -w

use strict;
use Benchmark;

timethese(1000000,
          {
           regex => sub {
                         my $wohnort = 'Dorfstrasse17';
                         my ($str,$hnr) = $wohnort =~ m/^([^0-9]+)([0-9]+)$/;
                        },
           splt1 => sub {
                         my $wohnort = 'Dorfstrasse17';
                         @kk = split(/[a-z]/i,$wohnort);
                         @kk = split(/[0-9]/,$wohnort);
                        },
           splt2 => sub {
                         my $wohnort = 'Dorfstrasse17';
                         my ($str,$hnr) = split(/\s*([0-9]+)/,$wohnort,2);
                        },
           splt3 => sub {
                         my $wohnort = 'Dorfstrasse17';
                         my ($str,$hnr) = split(/(?=[0-9])/, $wohnort, 2);
                        }
          }
         );

eof

H:\Perl\bin>perl bench.pl
Benchmark: timing 1000000 iterations of regex, splt1, splt2, splt3...
     regex:  6 wallclock secs ( 5.16 usr +  0.00 sys =  5.16 CPU) @ 193836.01/s (n=1000000)
     splt1: 15 wallclock secs (15.05 usr +  0.02 sys = 15.07 CPU) @ 66343.79/s (n=1000000)
     splt2: 10 wallclock secs ( 9.06 usr +  0.00 sys =  9.06 CPU) @ 110326.57/s (n=1000000)
     splt3:  6 wallclock secs ( 5.63 usr +  0.00 sys =  5.63 CPU) @ 177683.01/s (n=1000000)

H:\Perl\bin>