/^(?!.*(?:evil|bad|wrong))/
[download]

Bingo. This is the correct solution.

I tried this one a little earlier (knowing why it wasn't working):

/^.*(?!evil|bad|wrong)/
[download]

...but it never occurred to me to put the .* INSIDE the negative lookahead.

<slaps forehead/>

Thanks to the rest of you for helping, but most of you actually reversed the problem. The regex needed to return a match from the clean string, not from the string containing the blacklisted tokens.

Thanks!!

--Benji

Just for the sake of completeness, here's an alternative:

  /^(?:(?!evil|bad|wrong).)*\z/
[download]

Eimi's solution is much faster for the problem as given, but this approach gives more control - if your problem becomes any more complex in the future it's a useful trick to have available.

Hugo

use strict;
use warnings;

while(<DATA>) {
    print unless /(evil|bad|wrong)/i;
}

__DATA__
This string contains no blacklisted tokens.
This string is evil and wrong.
[download]

Updated: added i switch.

my $good = "this string contains no blacklisted tokens";
my $bad = "this string is evil and wrong";

my @blacklist = ('evil', 'bad', 'wrong');
my $re = join '|', @blacklist;

warn 'good string is ok' if $good !~ /\b(?:$re)\b/i;
warn 'bad string failed' if $bad =~ /\b(?:$re)\b/i;
[download]

my @blacklist = ('evil', 'bad', 'wrong');
my $a = "this string contains no blacklisted tokens";
my $b = "this string is evil and wrong";

my $regex = join '|', @blacklist;

# The !~ is a better way to do !(foo =~ /regex/)
if ($a !~ /$regex/) {
    print "Success with a!\n";
}
else {
    print "Failure with a\n";
}

if ($b !~ /$regex/) {
    print "Success with b!\n";
}
else {
    print "Failure with b\n";
}
[download]

Update: Sigh. And looking too much at the "match any string" bit caused me to reverse the logic of what you need. Must think on that a bit more...

Update the second:The problem with negative lookaheads is that it's easy to match when you don't want to. To simplify your problem a bit, just look at the word evil and try to match something without that in the text.

my $regex = '.(?!evil)';

if ("this is so evil" =~ /($regex)/) {
    print "yay: $1\n";
}
[download]

Prints "yay: t" for the simple fact that it matches at the beginning of the string. So you'll get false positives, because you need to check the whole string for the bad token, and negate that. I'd love to see a solution to this regex issue, personally, but I don't know of one.

Honestly I think the solution will not be a regex, but a change in the surrounding code. Even if it is possible to create a regex that does what you need, it's going to be more complex than making a few simple changes to the conditional. Reverse the logic of the if statement, the regex is trivial. Therefore future programmers won't curse your name and all will be happy in the world. Or something.

I'd be inclined to do that as a sub like this:

 sub is_blacklisted {
  my $string = $_[0];
  for (@blacklist){
    # add \b /i to re as required
    return 1 if ( $string =~ /$_/ );
  }
  return 0;
 }
[download]

<Rant> Where I am working at the moment they have a set of key words blocked on the proxy server. I discovered it by accident doing a search for "death and taxes", apparently "death" is not an acceptable topic even though it was completely benign in the context of my search, you can't even do a dictionary search for death! "Dead" on the other hand is ok...

I had a similar problem a couple of days ago and opted for the for solution. Are there any advantages either way for using this or a regex?

Also could you outline what you do with the return value on exiting the sub?

Thanks