Description:

I am trying to write a regular expression to aid in the parsing of a configuration file and I want to allow the users of the configuration file to specify any character in their directive including the special characters for the file format (comment (#), equivalance (=), etc.) and I have been trying to write an appropriate regex with no success. I am new at this and I have tried:

s/([^(?:([^\\]|\A)\\(\\{2})*\#)]*)(.*)/$1/
[download]

which represents a # not followed by an odd number of \'s (since \\# is the \ character metaquoted followed by a comment) but that didn't work becasue ^ only represents single characters and not sets of characters. I then tried the perl 5.005 negative lookbehind (?<!) but it only allows fixed width lookbehinds and I want to allow any number of \'s. Currently I am doing:

split /\Q#\E/;
$_ = $_[0];
if(/\A\s*\Z/) {
  next;
}
$string = $_;
for($i = 1; $i <= $#_; $i++) {
  $_ = $_[$i - 1];
  m/(.)((\\){2})*\Z/;
  if("$1" eq "\\") {
    $string .= "\#" . $_[$i];
  } else {
   last;
  }
}
[download]

Can anyone suggest a regex to do all that work? I remember seeing one to correctly parse a C string somewhere which would deal with these same issues, but hard as I look I cannot find it.

Will

s/(^[(?:([^\\|\A)\\(\\{2})*\#)]*)(.*)/$1/

(?<!)

$c = "\#";
$m = "\\";

while(<IN>) {
  chomp;
  split /\Q$c\E/;
  $_ = $_[0];
  next if(/\A\s*\Z/);
  $string = $_;
  for($i = 1; $i <= $#_; $i++) {
   $_ = $_[$i - 1];
   m/(.)((\Q$m\E){2})*\Z/;
   if("$1" eq "$m") {
     $string .= "$c" . $_[$i];
   } else {
     last;
   }
  }
}