Subversion Repositories vgwhois

#!/usr/bin/php

<?php

#

#  generic Whois - Automatic Pattern Generator: TLDs

#

#  (c) 2012-2015 Daniel Marschall, ViaThinkSoft [www.viathinksoft.de]

#

#  Distribution, usage etc. pp. regulated by the current version of GPL.

#

#

#  Version 2015-07-13

#

require_once __DIR__ . '/config.inc.php';

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

define('PATTERN_DIR', __DIR__ . '/../../main/pattern');

define('DOMAINS_PATTERN_FILE', PATTERN_DIR.'/domains');

define('CACHE_FILE_DIR', __DIR__ . '/../.cache/cache');

error_reporting(E_ALL | E_NOTICE | E_STRICT | E_DEPRECATED);

$iana_tld_data = null;

// Step 1:

// Check if in the meantime TLDs without an official whois server were updated to have one

// Attention/TODO: A change of the whois name still needs manual intervention!

$domains_cont_original = file_get_contents(DOMAINS_PATTERN_FILE);

$domains_cont_new = preg_replace_callback(

	'@# TODO: Entry generated automatically\. Needs manual check\.\n:notice\|\|Whois server unknown \((.*)\)\n\\\.(.*)\$\n@imU',

	function ($treffer) {

		$in_all = $treffer[0];

		$in_ts  = $treffer[1];

		$in_tld = $treffer[2];

		$days_passed = (time()-strtotime($treffer[1]))/(60*60*24);

		if ($days_passed < DOMAINS_RECHECK_MISSING_WHOIS_SERVERS) {

			return $in_all; // leave everything unchanged

		}

		$whois_serv = find_rootzone_whois_server($in_tld);

		if (!$whois_serv) {

			// Nothing found. Just update last check date.

			return str_replace($in_ts, date('Y-m-d'), $in_all);

		} else {

			// Update the entry

			return ":whois|$whois_serv\n.${in_tld}\$\n";

		}

	},

	$domains_cont_original

);

if ($domains_cont_original != $domains_cont_new) {

	file_put_contents(DOMAINS_PATTERN_FILE, $domains_cont_new);

	gwi_update_domains_patternfile();

}

// Step 2:

// Search for new TLDs which are not in our pattern file

if (!isset($iana_tld_data)) {

	$iana_tld_data = explode("\n", cached_file(IANA_TLD_REGISTRY, CACHE_FILE_DIR));

	$iana_tld_data = array_map('trim', $iana_tld_data);

}

foreach ($iana_tld_data as $tld) {

	if ($tld == '') continue;

	if ($tld[0] == '#') continue;

	$tld = strtolower($tld);

	if (!does_exist($tld)) {

		$whois_serv = find_rootzone_whois_server($tld);

		$to_append = "\n";

		if ($whois_serv) {

			$to_append .= ":whois|$whois_serv\n";

		} else {

			$to_append .= "# TODO: Entry generated automatically. Needs manual check.\n";

			$to_append .= ":notice||Whois server unknown (".date('Y-m-d').")\n";

		}

		$to_append .= "\\.$tld$\n";

		file_put_contents(DOMAINS_PATTERN_FILE, $to_append, FILE_APPEND);

		gwi_update_domains_patternfile();

		echo "Added: $tld\n";

	}

}

# ------------------------------------------------------

function does_exist($tld) {

	$cont = get_united_pattern();

	$cont = preg_replace('@#[^\n]*\n@ismU', '', $cont);

	$tld  = strtolower($tld);

	$cont = strtolower($cont);

	$cont = str_replace(array('(', ')'), '', $cont);

	return (strpos($cont, "\\.$tld\$\n") !== false);

}

function gwi_update_domains_patternfile() {

	$now = date('Ymd');

	$pcont_original = file_get_contents(DOMAINS_PATTERN_FILE);

	$pcont = $pcont_original;

	$count = 0;

	$pcont = preg_replace("@#: version (\\S+)@i", "#: version $now", $pcont, 1, $count);

	if ($count == 0) {

		// Add header

		$pcont = "#: version $now\n".

		         "# Domains\n".

		         "# This file can be updated by ".__DIR__."/generate_domains (only additions of new entries)\n".

		         "# --------------------------------------------------------------------\n".

		         "\n".$pcont;

	}

	if ($pcont != $pcont_original) {

		file_put_contents(DOMAINS_PATTERN_FILE, $pcont);

	}

}

function find_rootzone_whois_server($tld) {

	$whois_serv = iana_get_rootzone_whois_server($tld);

	// Try to find undocumented whois servers

	if (TRY_FINDING_HIDDEN_WHOIS_SERVERS) {

		// TODO: also try out to use the URL of the homepage (in IANAs root DB)

		if (!$whois_serv) {

			$check_server = "whois.nic.$tld";

			if (gwitc_is_port_open($check_server, 43)) {

				$whois_serv = $check_server;

			}

		}

		if (!$whois_serv) {

			$check_server = "whois.$tld";

			if (gwitc_is_port_open($check_server, 43)) {

				$whois_serv = $check_server;

			}

		}

		if (!$whois_serv) {

			$check_server = "$tld";

			if (gwitc_is_port_open($check_server, 43)) {

				$whois_serv = $check_server;

			}

		}

	}

	return $whois_serv;

}

function iana_get_rootzone_whois_server($tld) {

	$tld  = strtolower($tld);

	$cont = QueryWhoisServer('whois.iana.org', $tld);

	if (!preg_match('@whois:\\s*(\\S+)@i', $cont, $m)) return false;

	return $m[1];

}