Subversion Repositories vgwhois

#!/usr/bin/php
<?php

#
#  VGWhoIs (ViaThinkSoft Global WhoIs, a fork of generic Whois / gwhois)
#  Maintenance / Developer utilities
#
#  (c) 2012-2019 by Daniel Marschall, ViaThinkSoft <info@daniel-marschall.de>
#
#  License: https://www.gnu.org/licenses/gpl-2.0.html (GPL version 2)
#

#mb_internal_encoding('utf-8');

require_once __DIR__ . '/../../../shared/php_includes/common_functions.inc.php';
require_once __DIR__ . '/../../../shared/php_includes/idna_convert.class.php';

$punycoder = new idna_convert();

$in = '';
while($f = fgets(STDIN)) {
        $in .= $f;
}

$pre  = generateRandomToken($in);
$post = generateRandomToken($in);
$plc1 = generateRandomToken($in);

// to avoid that "domain......ati.tn" will be highlighted completely.
$in = preg_replace('@\\.{2,}+@u', '<<<'.$plc1.':${0}>>>', $in);

if ($argc == 1) {
        $out = array();
        exec(__DIR__ . '/allpatterns', $out, $code);
        if ($code != 0) {
                echo __DIR__ . "/allpatterns failed with code $code\n"; # TODO: STDERR
                exit(1);
        }
        foreach ($out as $o) {
                $o = trim($o);
                if ($o == '') continue;
                $argv[] = $o;
        }
}

array_shift($argv);
# $i = 0;
foreach ($argv as $x) {
        # TODO: problem: $x darf nicht quoted werden, darf aber auch nicht @ enthalten -> dynamisch rausfinden welcher delimiter nicht vorkommt
        # TODO: there are several problems using this mechanism. It is very hard to highlight everything which is queryable through vgwhois.
        #       for example: (1) it is not possible to use \S+ in the pattern files, because it would match everything in the vgwhois output, even though it was only meant to match everything in the query
        #                    (2) if the rule .com.xx is applied after .xx , the .com.xx will not match, since .xx is already wrapped with $pre and $post

#       $i++;
#       $pre = "[start$i]";
#       $post = "[end$i]";

        $in = preg_replace($x, $pre.'${0}'.$post, $in);

        $c = 0;

        $x = preg_replace_callback("@(xn\-\-[0-9a-z\-]+)@ui", "punycodeCB", $x, -1, $c);
        if ($c > 0) {
                $in = preg_replace($x, $pre.'${0}'.$post, $in);
        }
}

# Remove nesting (( x (( y )) a )) -> (( x a ))
do {
        $ok = true;
        $in = preg_replace_callback("@$pre(.*$post)@Uu", "verschachtelungCB", $in);
} while (!$ok);

# colorize
$in = str_replace($pre,  "\033[41m\033[37m", $in);
$in = str_replace($post, "\033[0m", $in);

# undo it again (see above)
$in = preg_replace('@<<<'.$plc1.':([^>]+)>>>@u', '${1}', $in);

echo $in;

# ---

function verschachtelungCB($treffer) {
        global $pre, $post, $ok;

        if (strpos($treffer[1], $pre) === false) {
                return $pre.$treffer[1];
        } else {
                $ok = false;
                return $pre.str_replace(array($pre,$post), '', $treffer[1]);
        }
}

function punycodeCB($treffer) {
        global $punycoder;

        $out = "";
        $x = $punycoder->decode(strtolower($treffer[0]));
        preg_match_all('/./u', $x, $results);
        foreach ($results[0] as $m) {
                $out .= '\\x{'.dechex(_uniord($m)).'}';
        }

        return $out;
}