s///e

Interesting...

This might not be the most efficient way to do it, but you could do:

#!/usr/bin/perl -w

use strict;

while(<>) {
         s/(.*?)(....)$/("X" x length($1)).$2/e;
         print;
}
[download]

The efficient way is probably:

#!/usr/bin/perl -w

use strict;

while(<>) {
    chomp;
    if(length > 4) {
         my $tail = substr($_,-4);
         $_=('X'x (length($_)-4)).$tail;
    }
    print "$_\n";
}
[download]

Of course, I could be wrong; the regex could be faster than substr. I'll defer to someone with enough tuits to run a benchmark :)
--
Mike

Edit: maverick (++) had the tuits; substr wins by a large margin.

Comment on s///e Select or Download Code

Replies are listed 'Best First'.
Re: s///e by maverick (Curate) on May 02, 2002 at 19:25 UTC
Here's yer tuit... `use Benchmark; my $string = "1234567890"; timethese(-5, { 'Regexp' => sub { $string =~ s/(.?)(....)$/("X" +x length($1)).$2/e; }, 'Regexp D&P' => sub { $string =~ s/.(?!.{0,3}$)/x/g; +}, 'Substring' => sub { $string = length substr($string +,0,-4). substr($string,-4) } });` [download] Yields: `Benchmark: running Regexp, Regexp D&P, Substring, each for at least 5 +CPU seconds... Regexp: 6 wallclock secs ( 5.38 usr + 0.00 sys = 5.38 CPU) @ 17 +3245.35/s (n=932060) Regexp D&P: 6 wallclock secs ( 5.30 usr + 0.00 sys = 5.30 CPU) @ 20 +0480.57/s (n=1062547) Substring: 7 wallclock secs ( 5.45 usr + 0.00 sys = 5.45 CPU) @ 10 +04355.05/s (n=5473735)` [download] This is on a 2 GHz Pentium 4. Let's say we increase $string to 1000 characters... Benchmark: running Regexp, Regexp D&P, Substring, each for at least 5 CPU seconds... Regexp: 6 wallclock secs ( 5.29 usr + 0.00 sys = 5.29 CPU) @ 4194.33/s (n=22188) Regexp D&P: 5 wallclock secs ( 5.26 usr + 0.00 sys = 5.26 CPU) @ 2068.25/s (n=10879) Substring: 7 wallclock secs ( 5.26 usr + 0.00 sys = 5.26 CPU) @ 1022051.33/s (n=5375990) Update:* Added Dog_and_Pony's solution... Update 2:Fixed stupid bug in Substring code. , and . do different things... Update 3:abstracts spotted that my substring function has a cut and paste error (sheesh). jsprat's benchmark is accurate. The four arg substring seems to be the fastest. /\/\averick OmG! They killed tilly! You bleep!!	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re: s///e
by maverick (Curate) on May 02, 2002 at 19:25 UTC

use Benchmark;

my $string = "1234567890";

timethese(-5, {
                'Regexp'      => sub { $string =~ s/(.*?)(....)$/("X" 
+x length($1)).$2/e; },
                'Regexp D&P'  => sub { $string =~ s/.(?!.{0,3}$)/x/g; 
+},
                'Substring'   => sub { $string = length substr($string
+,0,-4). substr($string,-4) }
              });
[download]

Benchmark: running Regexp, Regexp D&P, Substring, each for at least 5 
+CPU seconds...
    Regexp:  6 wallclock secs ( 5.38 usr +  0.00 sys =  5.38 CPU) @ 17
+3245.35/s (n=932060)
Regexp D&P:  6 wallclock secs ( 5.30 usr +  0.00 sys =  5.30 CPU) @ 20
+0480.57/s (n=1062547)
 Substring:  7 wallclock secs ( 5.45 usr +  0.00 sys =  5.45 CPU) @ 10
+04355.05/s (n=5473735)
[download]

Benchmark: running Regexp, Regexp D&P, Substring, each for at least 5 CPU seconds...
    Regexp:  6 wallclock secs ( 5.29 usr +  0.00 sys =  5.29 CPU) @ 4194.33/s (n=22188)
Regexp D&P:  5 wallclock secs ( 5.26 usr +  0.00 sys =  5.26 CPU) @ 2068.25/s (n=10879)
 Substring:  7 wallclock secs ( 5.26 usr +  0.00 sys =  5.26 CPU) @ 1022051.33/s (n=5375990)

Update:

Dog_and_Pony

Update 2:

Update 3:

abstracts

jsprat

/\/\averick
OmG! They killed tilly! You *bleep*!!

[reply]
[d/l]
[select]