Abigail's extended regular expression is also an opportunity to show you a nifty module called YAPE::Regex::Explain.

ladoix% cat 290992
#!/usr/bin/perl

use YAPE::Regex::Explain;
print YAPE::Regex::Explain->new(qr/_test(?>\s+)(?!<)/)->explain;


ladoix% perl 290992
The regular expression:

(?-imsx:_test(?>\s+)(?!<))

matches as follows:

NODE                     EXPLANATION
----------------------------------------------------------------------
(?-imsx:                 group, but do not capture (case-sensitive)
                         (with ^ and $ matching normally) (with . not
                         matching \n) (matching whitespace and #
                         normally):
----------------------------------------------------------------------
  _test                    '_test'
----------------------------------------------------------------------
  (?>                      match (and do not backtrack afterwards):
----------------------------------------------------------------------
    \s+                      whitespace (\n, \r, \t, \f, and " ") (1
                             or more times (matching the most amount
                             possible))
----------------------------------------------------------------------
  )                        end of look-ahead
----------------------------------------------------------------------
  (?!                      look ahead to see if there is not:
----------------------------------------------------------------------
    <                        '<'
----------------------------------------------------------------------
  )                        end of look-ahead
----------------------------------------------------------------------
)                        end of grouping
----------------------------------------------------------------------
[download]

--
Allolex

Unfortunally, it only explains what it does, but it doesn't explain why it does so. Perhaps the most subtle part of the regex is (?>\s+).

Can you explain why it uses "no backtracking"? ;-)

Abigail

I would love to know why you did what you did with that regex. There are probably a lot of monks in the Monastery who could learn something from you and just need a little push in the right direction. :)

--
Allolex

I'm sure you're familiar with the "cut" operator, Abigail, but most of the other people here will find it largely underdocumented. So I dare to feel free to point towards the draft of a book on regular expressions that the same perlmonk is writing, who also wrote YAPE::Regex::Explain (actually, he wrote the whole YAPE suite).

I personally like the draft of book very much, and so, I like to plug it whenever I can. So here's the URL: http://japhy.perlmonk.org/book/. Check out chapter 8 for the "cut" operator — every chapter is a separate download, roughly around 12 pages each, either in MS Word or in PDF format. Recommended.

use Benchmark;
$str1 = "_test (folloed by 1 or more spaces)";
$str2 = "_test < xxx >";

timethese ( 1000000,
            { 'p1' => '&p1;',
              'p2' => '&p2;',
              'p3' => '&p3;',
              'p4' => '&p4;',
             } );

sub p1 ()
{
    $str1 =~ /_test(?>\s+)(?!<)/;
}

sub p2 ()
{
    $str1 =~ /_test(?:\s+)(?!<)/;
}

sub p3 ()
{
    $str2 =~ /_test(?>\s+)(?!<)/;
}

sub p4 ()
{
    $str2 =~ /_test(?:\s+)(?!<)/;
}
[download]

Benchmark: timing 1000000 iterations of p1, p2, p3, p4...
        p1:  3 wallclock secs ( 3.00 usr +  0.00 sys =  3.00 CPU)
             @ 333333.33/s (n=1000000)
        p2:  3 wallclock secs ( 2.79 usr +  0.00 sys =  2.79 CPU)
             @ 358422.94/s (n=1000000)
        p3:  3 wallclock secs ( 3.09 usr +  0.00 sys =  3.09 CPU)
             @ 323624.60/s (n=1000000)
        p4:  3 wallclock secs ( 2.82 usr +  0.00 sys =  2.82 CPU)
             @ 354609.93/s (n=1000000)
[download]

Considering that /_test(?:\s+)(?!<)/ is wrong, as demonstrated elsewhere in this thread, I fail to see your point.

Abigail

Hi Ab, but I have tested the following code, which were giving me the same test results on many test cases:


if ($str =~ /_test(?>\s+)(?!<)/) { print "test ok\n" }
if ($str =~ /_test(?:\s+)(?!<)/) { print "test ok\n" }
[download]

Could you please tell me as why the /_test(?:\s+)(?!<)/ is wrong? I want to learn. Thanks!

use Benchmark;
$str1 = "_test (folloed by 1 or more spaces)";
$str2 = "_test < xxx >";

timethese ( 1000000,
            { 'p1' => '&p1;',
              'p2' => '&p2;',
              'p3' => '&p3;',
              'p4' => '&p4;',
             } );

sub p1 ()
{
    $str1 =~ /_test(?>\s+)(?!<)/;
}

sub p2 ()
{
    $str1 =~ /_test(?:\s+)(?!<)/;
}

sub p3 ()
{
    $str2 =~ /_test(?>\s+)(?!<)/;
}

sub p4 ()
{
    $str2 =~ /_test(?:\s+)(?!<)/;
}
[download]

Benchmark: timing 1000000 iterations of p1, p2, p3, p4...
        p1:  3 wallclock secs ( 3.00 usr +  0.00 sys =  3.00 CPU)
             @ 333333.33/s (n=1000000)
        p2:  3 wallclock secs ( 2.79 usr +  0.00 sys =  2.79 CPU)
             @ 358422.94/s (n=1000000)
        p3:  3 wallclock secs ( 3.09 usr +  0.00 sys =  3.09 CPU)
             @ 323624.60/s (n=1000000)
        p4:  3 wallclock secs ( 2.82 usr +  0.00 sys =  2.82 CPU)
             @ 354609.93/s (n=1000000)
[download]

Switching timethese with cmpthese , here's the math

Win32  ActivePerl 5.6.1 (Build 633)
       Rate  p4  p3  p1  p2
p4 865052/s  -- -1% -1% -4%
p3 876424/s  1%  -- -0% -3%
p1 877193/s  1%  0%  -- -3%
p2 901713/s  4%  3%  3%  --

Win32  ActivePerl 5.8.0 (build 804)

       Rate  p3  p1  p2  p4
p3 831255/s  -- -3% -5% -8%
p1 853971/s  3%  -- -3% -5%
p2 876424/s  5%  3%  -- -3%
p4 900901/s  8%  5%  3%  --

p1 and p3 use the "cut" operator. The optimization depends on your perl version.