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

Can anyone tell me why this pattern is never matched when parsing a file with commented strings in it? 
            my $word = '#';
            my $next = "1.0";
            while (my $from = $t->search(-regexp, "\\b$word\\b", $next
+, "end")) {
               my $word_len = length $' + 1;
               print $word;
               print $word_len;
               $next = "$from + $word_len chars";
               $t->tagAdd("red", $from, $next);
               $t->tagAdd("bold", $from, $next);
            }

[download]
It works fine for ever other word I have being searhed for but not the # character.

Replies are listed 'Best First'.
Re: regexp pattern match help?
by Roger (Parson) on Dec 03, 2003 at 00:47 UTC
    It's because you had \b switches to match a word boundary, where # is not considered as a word, but rather a boundary instead.

    Update: The following code is flawed, see Abigail-II's comment below.

    If you want to match '#' in your regex, you could do this instead -
    my $str = "This is a line # with comment";
my $word = '#';
while ($str =~ /[^\B#]($word)[^\B#]/g) {
    print "$1\n";
}
__OUTPUT__
#

    [download]
    Notice the [^\B#] idiom, what it means is that I want a character set of \B, non-word boundary, and #, and then take the compliment of the set. So the result will be a word boundary that does not match on the # character.

    Update: Thanks to Abigail-II for the detailed analysis of [^\B#].

    Ok, below is one way I think would fix the problem -
    my $str = "This is a line # with comment Boss.";
my $word = "#";

# define custom \b
my $b = qr/(?:(?=\S)(?<!\S)|(?!\S)(?<=\S))/;

# and match on non-space characters
while ($str =~ /$b(\S+)$b/g) {
    print "$1\n";
}

# or ignore the boundaries completely and match on non-space character
+s
while ($str =~ /(\S+)/g) {
    print "$1\n";
}

    [download]
[reply]
[d/l]
[select]
      \B and \b are zero-width assertions, and therefore, they don't make any sense inside a character class. Hence, [^\B#] doesn't do what you think: 
      $ perl -Dr -ce '/[^\B#]/'
Compiling REx `[^\B#]'
size 12 Got 100 bytes for offset annotations.
first at 1
   1: ANYOF[\0-"$-AC-\377{unicode_all}](12)
  12: END(0)
stclass `ANYOF[\0-"$-AC-\377{unicode_all}]' minlen 1 
Offsets: [12]
        1[6] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 7[0] 
Omitting $` $& $' support.

EXECUTING...

-e syntax OK
Freeing REx: `"[^\\B#]"'
$

      [download]
      Which means that [^\B#] matches any character that is not a B nor a #.

      Abigail

[reply]
[d/l]
[select]
        Ok cool, good to know. I think I have a decent way of accomplishing this but need a good way to set the length of string to be colored to the whole commented string. Here is what I have so far: 
                    my $word = '#';
            my $next = "1.0";
            while (my $from = $t->search(-regexp, "\\B$word\\B", $next
+, "end")) {
               my @comment = split(/#/, $_);
               #print "\$comment[0] equals ",$comment[0],"\n";
               #print "\$comment[1] equals ",$comment[1],"\n";
               if ($comment[1]) {
                  my $word_len = length $comment[1] + length $word;
               }else{
                  my $word_len = length $comment[0] + length $word;
               }
               print $word;
               print $word_len;
               $next = "$from + $word_len chars";
               $t->tagAdd("orange", $from, $next);
               $t->tagAdd("bold", $from, $next);
            }

        [download]
        I decided to use split to accomplish what I wanted but for some reason on each string the length only ends up being 5 so the first 5 characters of the commented strings gets colored. How can I color the whole comment once the comment character is found. Oh and am I searching for the comment symbol the best way using the "B"?
[reply]
[d/l]

Back to Seekers of Perl Wisdom