From: eggie5 <eggie5@...>
Date: 2007-09-26T02:50:16+09:00
Subject: Re: Extract numbers from string

On Sep 25, 1:31 am, "Jes�s Gabriel y Gal�n" <jgabrielyga...@gmail.com>
wrote:
> On 9/25/07, Pe�a, Botp <b...@delmonte-phil.com> wrote:
>
> > From: David A. Black [mailto:dbl...@rubypal.com]
>
> > #Don't forget \D:
> > #
> > #   old_num.gsub(/\D/, "")
>
> I always wanted try the benchmark stuff, and this seemed like a good
> opportunity :-).
>
> cat remove_nondigits.rb && ruby remove_nondigits.rb
> # remove_nondigits.rb
> # 25 September 2007
> #
>
> require 'benchmark'
>
> n = 100_000
> old_num = "(555) 55-555-55"
>
> Benchmark.bmbm do |x|
>     x.report("scan") do
>       n.times {old_num.scan(/\d/).join}
>     end
>     x.report("gsub") do
>       n.times {old_num.gsub(/[^\d]/,"")}
>     end
>     x.report("gsub2") do
>       n.times {old_num.gsub(/\D/, "")}
>     end
>     x.report("tr") do
>       n.times {old_num.tr("^0-9","")}
>     end
>     x.report("delete") do
>       n.times {old_num.delete("^0-9")}
>     end
>     x.report("split") do
>       n.times {old_num.split(/[^\d]/).join}
>     end
> end
> Rehearsal ------------------------------------------
> scan     1.080000   0.000000   1.080000 (  1.144654)
> gsub     0.380000   0.010000   0.390000 (  0.403057)
> gsub2    0.390000   0.010000   0.400000 (  0.416074)
> tr       0.290000   0.020000   0.310000 (  0.320764)
> delete   0.270000   0.010000   0.280000 (  0.294078)
> split    0.700000   0.000000   0.700000 (  0.714139)
> --------------------------------- total: 3.160000sec
>
>              user     system      total        real
> scan     1.080000   0.020000   1.100000 (  1.101550)
> gsub     0.380000   0.010000   0.390000 (  0.404968)
> gsub2    0.380000   0.010000   0.390000 (  0.409364)
> tr       0.290000   0.020000   0.310000 (  0.318723)
> delete   0.280000   0.010000   0.290000 (  0.294650)
> split    0.700000   0.000000   0.700000 (  0.713261)
>
> So it seems that in terms of speed, delete wins, second place for tr,
> and in the fight between [^\d] and \D the first one wins by a little
> (very little) margin :-).
>
> Cheers,
>
> Jesus.

That's awesome, thanks.