Regex word boundary and escaped characters

[ryantate]

The regex word boundary meta character, \b, does not seem to match start of string if the first word character of the string is escaped. Is this expected behavior?

Here is an example of the behavior in question:

apocalypse.OCF [178] perl5.8.5 -e 'my $s = "\@testing"; 
$s =~ s/\b\@testing\b//i; print "$s\n";' 

@testing

apocalypse.OCF [179] perl5.8.5 -e 'my $s = "+testing"; 
$s =~ s/\b\+testing\b//i; print "$s\n";' 

+testing

apocalypse.OCF [180] perl5.8.5 -e 'my $s = "testing"; 
$s =~ s/\btesting\b//i; print "$s\n";' 


apocalypse.OCF [181]

For now I'm using (\b|^)\@testing(\b|$) as a workaround, but this behavior seems at odds with perlre, which states:

A word boundary ( \b ) is defined as a spot between two characters that has a \w on one side of it and and a \W on the other side of it (in either order), counting the imaginary characters off the beginning and end of the string as matching a \W .

Is this a bug?

Update: No, of course, it's not a bug. @ and + being non word chars, there is no \w char to match. Duh. ;--> Thanks for the many quick responses.

[tye]

\b only matches the start of the string if the first character is a \w character because it "[counts] the imaginary characters off the beginning and end of the string as matching a \W". That is, "+testing" is treated as if it were the middle of "*+testing*" (using "*" as the "imaginary characters off the beginning and end of the string") and there is no \b between the "*" and the "+".

[Paladin]

No, it's doing exactly as it states. @ matches \W and the start of the string also "matches" \W, so there is no \b between them. Same goes for the string starting with +. In the third example, t matches \w so the \b matches the space between the start of the string (\W) and the t (\w)

[dave_the_m]

counting the imaginary characters off the beginning and end of the string as matching a \W

You've got a \W (start of string) directly followed by a another \W (@ or +) so of course \b isn't going to match there.

Dave.

[Roy Johnson]

This is a case where you'd roll your own lookahead and lookbehind, rather than using the ready-made \b. Perhaps @testing that is not preceded or followed by non-space:

s/(?<!\S)\@testing(?!\S)//i;

[Eimi Metamorphoumai]

Others have answered your question, but I'd just like to point out that if you put a \b next to a non-word character, you're likely to get the exact opposite of what you want. That is, it will only match if the character next to it is a word character. So another way to fix it would be to use \B to assert that the character before you is also a non-word character.

#!/usr/bin/perl -l

use strict;
use warnings;

while(<DATA>){
  chomp;
  print;
  print /\b\@testing\b/ ? "Matched \\b" : "Did not match \\b";
  print /\B\@testing\b/ ? "Matched \\B" : "Did not match \\B";
  print "";
}

__DATA__
@testing
foo@testing
foo @testing