#!perl -w

=head1 NAME

uribl - URIBL blocking plugin for qpsmtpd

=head1 DESCRIPTION

This plugin implements DNSBL lookups for URIs found in spam, such as that
implemented by SURBL (see E<lt>http://surbl.org/E<gt>).  Incoming messages are
scanned for URIs, which are then checked against one or more URIBLs in a
fashion similar to DNSBL systems.

=head1 CONFIGURATION

To enable the plugin, add it to I<~qpsmtpd/config/plugins>.  The list of
URIBLs to check should be placed in I<uribl_zones> in the config directory
(typically I<~qpsmtpd/config>).

The format of the I<uribl_zones> file is a list of URIBL DNS zones, one per
line, consisting of one or more columns separated by whitespace.  The first
column (the only mandatoy one) should consist of the URIBL zone.

The second column may contain a comma-delimited list of integers selecting
mask values to be applied to the A record(s) returned from a URIBL.  This
enables the use of composite DNSBLs, such as multi.surbl.org, where several
lists are combined so they may be accessed with a single query; any returns
are checked against the mask of lists you're interested in.  If unspecified,
or if a negative number is given, all lists in a composite URIBL will be
checked.  URIBL operators prefer that you use the composite lists to reduce
their own query load, and it's more efficient for qpsmtpd as well.

The third column specifies an action, which overrides the default action
configured with the I<action> setting discussed below.

For example:

=over 4

    multi.surbl.org    2,8    deny
    ob.surbl.org    1    add-header

=back

You may specify the following config option(s) in the I<qpsmtpd/config> file:

=over 4

=item action

Specifies what to do when a URI is matched in a URIBL.  Available options are
I<add-header> (the default) I<deny> and I<denysoft>.  If set to add-header, an
X-URIBL-Match: header will be added explaining the URIBL entry found.  If set
to 'deny,' the delivery will be declined with a hard failure.  If set to
denysoft, the delivery will be soft failed (this is probably not a good idea.)

=item timeout

Timeout for DNS requests, in seconds.  The default is 30 seconds.  DNS
requests are issued asynchronously and in parallel for all hosts found
in URIs in the mail; the same timeout will apply to each; see the
Net::DNS documentation for details.

=item scan-headers

If set true, any headers found in the URIs will be checked as well.  Disabled
by default.

=back

=head1 CAUTIONS

When used in I<deny> or I<denysoft> mode, a URIBL check can block not
only the original spam containing a listed URI, but mail unintentionally
carrying that URI, such as forwarded complaints.  The uribl checks should
only be used in these modes if you know what you're doing.

The URI scanner used by the uribl plugin is quite aggressive, and attempts to
detect all forms of URIs supported by typical MUAs (even those that lack a
protocol specification, for example.)  However, it does not attempt to detect
URIs that have been mangled beyond programmatic reconstruction.  Even so, it
may issue spurious lookups on unintended URIs, especially those in non-text
sections of the mail.

=head1 COPYRIGHT

uribl is copyright 2004-2007 by Devin Carraway E<lt>qpsmtpd@devin.comE<gt>.  It
may be used and redistributed under the same terms as qpsmtpd itself.

=cut

use Net::DNS::Resolver;
use Time::HiRes qw(time);
use IO::Select;

use Qpsmtpd::Constants;

use strict;
use warnings;

# ccTLDs that allocate domain names within a strict two-level hierarchy,
# as in *.co.uk
my %strict_twolevel_cctlds = (
    'ac' => 1,
    'ae' => 1,
    'uk' => 1,
    'ai' => 1,
    'ar' => 1,
    'at' => 1,
    'au' => 1,
    'az' => 1,
    'bb' => 1,
    'bh' => 1,
    'bm' => 1,
    'br' => 1,
    'bs' => 1,
    'ca' => 1,
    'ck' => 1,
    'cn' => 1,
    'co' => 1,
    'cr' => 1,
    'cu' => 1,
    'cy' => 1,
    'do' => 1,
    'et' => 1,
    'ge' => 1,
    'hk' => 1,
    'id' => 1,
    'il' => 1,
    'jp' => 1,
    'kr' => 1,
    'kw' => 1,
    'lv' => 1,
    'sg' => 1,
    'za' => 1,
);

# async version: OK
sub init {
    my ($self, $qp, %args) = @_;

    $self->{action} = $args{action} || 'add-header';
    $self->{timeout} = $args{timeout} || 30;
    # scan-headers was the originally documented name for this option, while
    # check-headers actually implements it, so tolerate both.
    $self->{check_headers} = $args{'check-headers'} || $args{'scan-headers'};

    $args{mask} ||= 0x00ffffff;
    $self->{mask} = 0;

    my @zones = $self->qp->config('uribl_zones');
    for (@zones) {
        chomp;
        next if !$_ or /^\s*#/;
        my @z = split (/\s+/, $_);
        next unless $z[0];

        my $mask = 0;
        $z[1] ||= 0x00ffffff;
        for (split /,/, $z[1]) {
            unless (/^(-?\d+)$/) {
                $self->log(LOGERROR, "Malformed mask $_ for $z[0]");
                return undef;
            }
            $mask |= $1 >= 0 ? $1 : 0x00ffffff;
        }
        my $action = $z[2] || $self->{action};
        unless ($action =~ /^(add-header|deny|denysoft|log)$/) {
            $self->log(LOGERROR, "Unknown action $action for $z[0]");
            return undef;
        }

        $self->{uribl_zones}->{$z[0]} = {
            mask => $mask,
            action => $action,
        };
    }
    keys %{$self->{uribl_zones}} or return 0;

    my @whitelist = $self->qp->config('uribl_whitelist_domains');
    $self->{whitelist_zones} = {
        ( map { ($_ => 1) } @whitelist )
    };

    $self->init_resolver;
}

# async version: not used
sub register {
    my $self = shift;

    $self->register_hook('data_post', 'data_handler');
}

# async version: not used
sub send_query {
    my $self = shift;
    my $name = shift || return undef;
    my $count = 0;

    $self->{socket_select} ||= new IO::Select or return undef;
    for my $z (keys %{$self->{uribl_zones}}) {
        my ($s, $s1);
        my $index = {
            zone => $z,
            name => $name,
        };

        next unless $z;
        next if exists $self->{sockets}->{$z}->{$name};
        $s = $self->{resolver}->bgsend("$name.$z", 'A');
        if (defined $s) {
            $self->{sockets}->{$z}->{$name}->{'a'} = $s;
            $self->{socket_select}->add($s);
            $self->{socket_idx}->{"$s"} = $index;
            $count++;
        } else {
            $self->log(LOGERROR,
                       "Couldn't open socket for A record '$name.$z': ".
                       ($self->{resolver}->errorstring || 'unknown error'));
        }

        $s1 = $self->{resolver}->bgsend("$name.$z", 'TXT');
        if (defined $s1) {
            $self->{sockets}->{$z}->{$name}->{'txt'} = $s1;
            $self->{socket_select}->add($s1);
            $self->{socket_idx}->{"$s1"} = $index;
            $count++;
        } else {
            $self->log(LOGERROR,
                       "Couldn't open socket for TXT record '$name.$z': ".
                       ($self->{resolver}->errorstring || 'unknown error'));
        }

        $self->{sockets}->{$z}->{$name} = {};
    }
    $count;
}

# async version: not used
sub lookup_finish {
    my $self = shift;
    $self->{socket_idx} = {};
    $self->{sockets} = {};
    undef $self->{socket_select};
}

# async version: OK
sub evaluate {
    my $self = shift;
    my $zone = shift || return undef;
    my $a = shift || return undef;

    my $mask = $self->{uribl_zones}->{$zone}->{mask} || $self->{mask};
    $a =~ /^(\d{1,3})\.(\d{1,3})\.(\d{1,3})\.(\d{1,3})$/ or return undef;
    my $v = (($1 & 0xff) << 24) |
        (($2 & 0xff) << 16) |
        (($3 & 0xff) << 8) |
        ($4 & 0xff);
    return ($v & $mask);
}

# async version: OK
sub lookup_start {
    my ($self, $transaction, $start_query) = @_;

    my $l;
    my $queries = 0;
    my %pending;
    my @qp_continuations;

    $transaction->body_resetpos;
	# if we're not looking for URIs in the headers, read past that point
	# before starting to actually look for any
    while (!$self->{check_headers} and $l = $transaction->body_getline) {
        chomp $l;
        last if !$l;
    }
    while ($l = $transaction->body_getline) {
        chomp $l;

        if ($l =~ /(.*)=$/) {
            push @qp_continuations, $1;
        } elsif (@qp_continuations) {
            $l = join('', @qp_continuations, $l);
            @qp_continuations = ();
        }

        # Undo URI escape munging
        $l =~ s/[=%]([0-9A-Fa-f]{2,2})/chr(hex($1))/ge;
        # Undo HTML entity munging (e.g. in parameterized redirects)
        $l =~ s/&#(\d{2,3});?/chr($1)/ge;
        # Dodge inserted-semicolon munging
        $l =~ tr/;//d;

        while ($l =~ m{
            \w{3,16}:/+            # protocol
            (?:\S+@)?            # user/pass
            (\d{7,})            # raw-numeric IP
            (?::\d*)?([/?\s]|$)        # port, slash
                            #  or EOL
            }gx) {
            my @octets = (
                (($1 >> 24) & 0xff),
                (($1 >> 16) & 0xff),
                (($1 >> 8) & 0xff),
                ($1 & 0xff)
                );
            my $fwd = join('.', @octets);
            my $rev = join('.', reverse @octets);
            $self->log(LOGDEBUG, "uribl: matched pure-integer ipaddr $1 ($fwd)");
            unless (exists $pending{$rev}) {
                $queries += $start_query->($self, $rev);
                $pending{$rev} = 1;
            }
        }
        while ($l =~ m{
            \w{3,16}:/+                  # protocol
            (?:\S+@)?                    # user/pass
            (\d+|0[xX][0-9A-Fa-f]+)\.    # IP address
            (\d+|0[xX][0-9A-Fa-f]+)\.
            (\d+|0[xX][0-9A-Fa-f]+)\.
            (\d+|0[xX][0-9A-Fa-f]+)
            }gx) {
            my @octets = ($1,$2,$3,$4);
            # return any octal/hex octets in the IP addr back
            # to decimal form (e.g. http://0x7f.0.0.00001)
            for (0..$#octets) {
                $octets[$_] =~ s/^0([0-7]+)$/oct($1)/e;
                $octets[$_] =~ s/^0x([0-9a-fA-F]+)$/hex($1)/e;
            }
            my $fwd = join('.', @octets);
            my $rev = join('.', reverse @octets);
            $self->log(LOGDEBUG, "uribl: matched URI ipaddr $fwd");
            unless (exists $pending{$rev}) {
                $queries += $start_query->($self, $rev);
                $pending{$rev} = 1;
            }
        }
        while ($l =~ m{
            ((?:www\.)?                             # www?
             [a-zA-Z0-9][a-zA-Z0-9\-.]+\.           # hostname
             (?:aero|arpa|asia|biz|cat|com|coop|    # tld
                edu|gov|info|int|jobs|mil|mobi|
                museum|name|net|org|pro|tel|travel|
                [a-zA-Z]{2})
            )(?!\w)        
            }gix) {
            my $host = lc $1;
            my @host_domains = split /\./, $host;
            $self->log(LOGDEBUG, "uribl: matched 'www.' hostname $host");

            my $cutoff = exists
                $strict_twolevel_cctlds{$host_domains[$#host_domains]} ? 3 : 2;
            if (exists $self->{whitelist_zones}->{
		join('.', 
		     @host_domains[($#host_domains-$cutoff+1)..$#host_domains])}) {
                $self->log(LOGINFO, "Skipping whitelist URI domain '$host'");
            } else {
                while (@host_domains >= $cutoff) {
                    my $subhost = join('.', @host_domains);
                    unless (exists $pending{$subhost}) {
                        $self->log(LOGINFO, "URIBL: checking sub-host $subhost");
                        $queries += $start_query->($self, $subhost);
                        $pending{$subhost} = 1;
                    }
                    shift @host_domains;
                }
            }
        }
        while ($l =~ m{
            \w{3,16}:/+                 # protocol
            (?:\S+@)?                   # user/pass
            (
	     [a-zA-Z0-9][a-zA-Z0-9\-.]+\.           # hostname
	     (?:aero|arpa|asia|biz|cat|com|coop|    # tld
                edu|gov|info|int|jobs|mil|mobi|
                museum|name|net|org|pro|tel|travel|
                [a-zA-Z]{2})
            )
            }gix) {
            my $host = lc $1;
            my @host_domains = split /\./, $host;
            $self->log(LOGDEBUG, "uribl: matched full URI hostname $host");

            my $cutoff = exists
                $strict_twolevel_cctlds{$host_domains[$#host_domains]} ? 3 : 2;
            if (exists $self->{whitelist_zones}->{
                    join('.', @host_domains[($cutoff-1)..$#host_domains])}) {

                $self->log(LOGINFO, "Skipping whitelist URI domain '$host'");
            } else {
                while (@host_domains >= $cutoff) {
                    my $subhost = join('.', @host_domains);
                    unless (exists $pending{$subhost}) {
                        $self->log(LOGINFO, "URIBL: checking sub-host $subhost");
                        $queries += $start_query->($self, $subhost);
                        $pending{$subhost} = 1;
                    }
                    shift @host_domains;
                }
            }
        }
    }
    $transaction->body_resetpos;

    return $queries;
}

# async version: not used
sub collect_results {
    my ($self, $transaction) = @_;

    my $matches = 0;
    my $complete = 0;
    my $start_time = time;
    while ($self->{socket_select}->handles) {
        my $timeout = ($start_time + $self->{timeout}) - time;
        last if $timeout <= 0;

        my @ready = $self->{socket_select}->can_read($timeout);

        SOCK: for my $s (@ready) {
            $self->{socket_select}->remove($s);
            my $r = $self->{socket_idx}->{"$s"} or next SOCK;
            $self->log(LOGDEBUG, "from $r: socket $s: ".
                join(', ', map { "$_=$r->{$_}" } keys %{$r}));
            my $zone = $r->{zone};
            my $name = $r->{name};
            my $h = $self->{sockets}->{$zone}->{$name};
            my $packet = $self->{resolver}->bgread($s)
                or next SOCK;

            for my $a ($packet->answer) {
                if ($a->type eq 'TXT') {
                    $h->{txt} = $a->txtdata;
                }
                elsif ($a->type eq 'A') {
                    $h->{a} = $a->address;
                    if ($self->evaluate($zone, $h->{a})) {
                        $self->log(LOGDEBUG,
                            "match in $zone");
                        $h->{match} = 1;
                        $matches++;
                    }
                }
            }

            $complete++;
        }
    }
    my $elapsed = time - $start_time;
    $self->log(LOGINFO,
        sprintf("$complete lookup%s finished in %.2f sec (%d match%s)",
            $complete == 1 ? '' : 's', $elapsed,
            $matches, $matches == 1 ? '' : 'es'));

    my @matches = ();
    for my $z (keys %{$self->{sockets}}) {
        for my $n (keys %{$self->{sockets}->{$z}}) {
            my $h = $self->{sockets}->{$z}->{$n};
            next unless $h->{match};
            push @matches, {
                action =>
                    $self->{uribl_zones}->{$z}->{action},
                desc => "$n in $z: ".
                    ($h->{txt} || $h->{a}),
            };
        }
    }

    $self->lookup_finish;

    return \@matches;
}

# async version: not used
sub data_handler {
    my ($self, $transaction) = @_;

    return (DECLINED) if $self->is_immune();

    my $queries = $self->lookup_start($transaction, sub {
        my ($self, $name) = @_;
        return $self->send_query($name);
    });

    unless ($queries) {
        $self->log(LOGINFO, "No URIs found in mail");
        return DECLINED;
    }

    my $matches = $self->collect_results($transaction);
    for (@$matches) {
        $self->log(LOGWARN, $_->{desc});
        if ($_->{action} eq 'add-header') {
            $transaction->header->add('X-URIBL-Match', $_->{desc}, 0);
        } elsif ($_->{action} eq 'deny') {
            return (DENY, $_->{desc});
        } elsif ($_->{action} eq 'denysoft') {
            return (DENYSOFT, $_->{desc});
        }
    }
    return DECLINED;
}

# async version: not used
sub init_resolver {
    my $self = shift;

    $self->{resolver} = new Net::DNS::Resolver or return undef;
    $self->{resolver}->udp_timeout($self->{timeout});
}