regex whitespace quantifers

halak77 has asked for the wisdom of the Perl Monks concerning the following question:

I am doing something which seems very simple to me, yet the regex is not behaving as I had thought it would. The file I am parsing is a fstab, I am trying to ignore commented lines (begin with "#.") I also want to ignore any 'comment' lines which are preceded by whitespace. My example file

# This file is a comment and should not be used
 # This comment has one leading space
   # This comment has three leading spaces
/dev/mapper/centos-var_log /var/log                xfs     defaults   
+     1 2
 /dev/donstest  /expert         none    1 1 -rw,noglob # this is a fs
 #/dev/donstest22   /expert         none    1 1 -rw,noglob # this is a
+ fS
[download]

and code

 
#!/usr/bin/perl
use strict;
open (MYFILE, "./file.txt" ) or die "Could not open file:  $! \n";
while ( my $line = <MYFILE> )
{
    chomp $line;
    if  ( $line =~ /^\s*[^#].*/ )
    {
        print "NOT Comment: $line \n";
    }
    else
    {
        print "IS Comment: $line \n";
    }
}
[download]

My output:

IS Comment: # This file is a comment and should not be used
NOT Comment:  # This comment has one leading space
NOT Comment:    # This comment has three leading spaces
NOT Comment: /dev/mapper/centos-var_log /var/log                xfs   
+  defaults        1 2
NOT Comment:  /dev/donstest     /expert         none    1 1 -rw,noglob
+ # this is a fs
NOT Comment:  #/dev/donstest22   /expert         none    1 1 -rw,noglo
+b # this is a fS
[download]

What am I missing? As I understand /^\s*^#/ to mean "match any strings which may have zero of more whitespace characters from line begining followed by any character not a '#' the \s* should be greedy up to the first non-whitespace character. If I replace my regex with /^\s*^#\s ... it works as desired.

Comment on regex whitespace quantifers Select or Download Code

Replies are listed 'Best First'.

Re: regex whitespace quantifiers
by Athanasius (Archbishop) on Apr 24, 2015 at 12:33 UTC

Hello halak77, and welcome to the Monastery!

Yes, the \s* is greedy, but if no match is found then the regex engine backtracks. So it tries to match the line:

 # This comment has one leading space
[download]

with one space character, but fails because the following character (#) doesn’t match [^#]. So the engine backtracks and tries zero whitespace characters, followed by [^#], which succeeds because the first character (a space) is not a hash, and so matches [^#].

If I replace my regex with /^\s*[^#\s]... it works as desired.

Yes, because now the case where a space matches the no-hash character class is explicitly excluded.

On the working of the Perl regex engine, see the section “The Little Engine That /Could(n't)?/” in The Camel Book (4^th Edition, 2012), pages 241–246.

Hope that helps,

Athanasius <°(((>< contra mundum Iustus alius egestas vitae, eros Piratica,

[reply]
[d/l]
[select]

Re: regex whitespace quantifers
by QM (Parson) on Apr 24, 2015 at 11:22 UTC

Seems you should turn your logic around, and match the comment:

/^\s*#/
[download]

~~Your regex should have a plus instead of a star after \s:~~

/^\s+[^#].*/
[download]

And apparently <strike> doesn't span code blocks.

-QM
--
Quantum Mechanics: The dreams stuff is made of

[reply]
[d/l]
[select]

Re: regex whitespace quantifers
by kcott (Archbishop) on Apr 24, 2015 at 15:13 UTC

G'day halak77,

Welcome to the Monastery.

"/^\s*[^#].*/"

[^#] is a character class containing all characters except '#'. What you really want is a character class containing only '#'.

The first line doesn't match: (^) - line start OK; (\s*) - match zero whitespace OK; ([^#]) - '#' is a '#' FAIL

The second line does match: (^) - line start OK; (\s*) - match zero whitespace OK; ([^#]) - '' is NOT a '#' OK

I'll leave you to continue through the remaining lines of data.

I'll also point out that the '.*' (at the end of the regex) is superfluous. It will always match: either none, one or some of any character matched by '.'. You're only interested in the front of the string, anyway.

Here's my test (pm_1124511_comment_removal.pl):

#!/usr/bin/env perl

use strict;
use warnings;

my $comment_line_re = qr{ \A \s* [#] }x;

while (<DATA>) {
    print "@{[/$comment_line_re/ ? 'IS' : 'NOT']} Comment: $_";
}

__DATA__
# This file is a comment and should not be used
 # This comment has one leading space
   # This comment has three leading spaces
/dev/mapper/centos-var_log /var/log                xfs     defaults   
+     1 2
 /dev/donstest  /expert         none    1 1 -rw,noglob # this is a fs
 #/dev/donstest22   /expert         none    1 1 -rw,noglob # this is a
+ fS
[download]

Output:

$ pm_1124511_comment_removal.pl
IS Comment: # This file is a comment and should not be used
IS Comment:  # This comment has one leading space
IS Comment:    # This comment has three leading spaces
NOT Comment: /dev/mapper/centos-var_log /var/log                xfs   
+  defaults        1 2
NOT Comment:  /dev/donstest  /expert         none    1 1 -rw,noglob # 
+this is a fs
IS Comment:  #/dev/donstest22   /expert         none    1 1 -rw,noglob
+ # this is a fS
[download]

-- Ken

[reply]
[d/l]
[select]