package CGI::NoPoison; use strict; use Carp 'croak'; our $VERSION = '3.11'; #note that we are overriding AUTOLOADed methods in CGI.pm in this way #so we aren't having to use the "no warnings 'redefine'" pragma ;) sub CGI::FETCH { return $_[0] if $_[1] eq 'CGI'; return undef unless defined $_[0]->param($_[1]); #Instead of returning a null-byte packed list, introducing #potential security problems, why not instead return an #anonymous array and just dereference it later ? #return join("\0",$_[0]->param($_[1])); my @a = $_[0]->param($_[1]); if ( scalar( @a ) > 1 ) { return [$_[0]->param($_[1])]; # return anon-array if more than one element } else { return $_[0]->param($_[1]); # behave normally otherwise, so we have a true drop-in replacement } } #and if we're going to do THAT, well, we might as well sub CGI::SplitParam { my ($param) = @_; #my (@params) = split ("\0", $param); #my (@params) = @{$param}; # wait, why bother with this? #return (wantarray ? @params : $params[0]); return (wantarray ? @{$param} : @{$param}[0]);# when you can just do this, instead! :) } #and also should probably sub CGI::STORE { my $self = shift; my $tag = shift; # this should now be a reference to a named or an anonymous array # ala $vals = [ qw(list goes here) ]; or $vals = \@ary; my $vals = shift; #my @vals = index($vals,"\0")!=-1 ? split("\0",$vals) : $vals; croak "Value list not an array reference" unless ref($vals) eq 'ARRAY'; #$self->param(-name=>$tag,-value=>\@vals); $self->param(-name=>$tag,-value=>$vals);# look ma, it's *already* a reference! } 1; __END__ =pod =head1 NAME CGI::NoPoison - No Poison Null Byte in CGI->Vars =head1 SYNOPSIS use CGI; use CGI::NoPoison my $m = CGI->new(); $m->param( -name=>'amplifier', -value=>['nine', 'ten', 'up to eleven'], ); my %h = $m->Vars(); # look ma, no splitting on poison null-bytes ( '\0' )! print "$_ => ", join ", ", @{$h{$_}} for keys %h; print "This one goes ", ($m->param('amplifier'))[2]; =head1 DESCRIPTION Simplicity itself. Instead of using a null-byte to separate multi-valued fields why not just use what CGI.pm B uses to store the values internally? "What's that?", you ask? Why, it's an anonymous array, of course, like anyone sensible would use. cgi-lib.pl may have been fine years and years ago, but this now-archaic throwback no longer needs us to bow to its demands. (is anyone still actually using it? yikes.) This does, however change how you parse CGI->Vars() (as an anon-array, not a C<\0>-packed string) and also how you set params. B you can properly test for inserted null-bytes in a secure environment B taking advantage of the convenience of the Vars() function. =head1 USAGE Include the 'use CGI::NoPoison' only after you've already done 'use CGI' so that it can replace the AUTOLOAD routines with these replacement functions instead. (By the way, the CGI.pm internal functions that we replace are: CGI::SplitParam, CGI::STORE, and CGI::FETCH, not that you'd actually ever use these directly :) Then, all you have to do is remember that anywhere you would have previously used C<\0> to split on, or to string-pack, just take an array reference, or use an anonymous array instead. See the L module documentation for details. =head1 BUGS None so far. :) Well, this may actually be a pretty wonky way of replacing those functions in CGI.pm, but hey, it worked here. YMMV. :D =head1 SUPPORT Yer on yer own with this one. Hopefully Lincoln Stein will get around to adding this as a -nopoison pragma to CGI.pm at some point. =head1 AUTHOR Scott R. Godin CPAN ID: SGODIN Laughing Dragon Services nospam@webdragon.net http://www.webdragon.net/ =head1 COPYRIGHT This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself. The full text of the license can be found in the LICENSE file included with this module. =head1 SEE ALSO L ( or message-id <20050209020155.15512.qmail@lists.develooper.com> ) Google around for "poison null byte" L, L =cut