Subversion Repositories vgwhois

#!/usr/bin/php

<?php

#

#  generic Whois - Automatic Pattern Generator: QA Monitor

#

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

#

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

#

#

#  Version 2015-04-27

#

# TODO: strikte trennung zwischen pattern entwicklung ( = im sdk enthalten ) und lokaler gwi entwicklung ( = packages, eigene vwi services )

# TODO: alles modular machen, vieles cachen

# TODO: was wenn ein rwi entfernt wird? prüfung notwendig dann!

#       --> das wird im whois-ping test geprueft

# TODO: auch unnoetige excludes (anormalities) in config.inc.php melden

# TODO pattern.d/tld --> check if all existing and disallocated TLDs are inside

# TODO: konsolenfarben z.b. grün für qa-audit

error_reporting(E_ALL | E_NOTICE | E_STRICT | E_DEPRECATED);

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

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

# ---

// This is like __DIR__, but does not resolve symlinks

// (useful for sharing StatMon source codes locally by symlinking)

define('DIR', dirname($_SERVER['SCRIPT_FILENAME']));

$anormale_whois = array();

require DIR . '/config.inc.php';

$pattern_cont = get_united_pattern();

$global_status = 0;

echo "* Checking for died whois servers and check if presumed dead whois servers (= already removed from the pattern file) have resurrected...\n";

exec(__DIR__ . "/whois-ping/whoisping", $out, $ec);

$loc_good = $ec == 0;

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	echo "[ FAIL ]\n";

	$global_status=1;

}

# ---

echo "* Checking the testcases in background mode...\n";

exec(__DIR__ . "/testcases/batch --mode b", $out, $ec);

$loc_good = $ec == 0;

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	echo "[ FAIL ]\n";

	$global_status=1;

}

# ---

echo "* Check if every deleted TLD has a notice ( ".DELETED_TLD_LIST." )\n";

# does not exist as static file at VTS server

$del_tld_cont = explode("\n", cached_file(DELETED_TLD_LIST, CACHE_FILE_DIR));

$loc_good = true;

foreach ($del_tld_cont as $tld) {

	$tld = trim($tld);

	if ($tld == '') continue;

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

	$tld = strtolower($tld);

	if (!preg_match('@:notice\\|\\|Disallocated by IANA([^\\n]*)[\\r\\n]+\\\\.'.$tld.'\\$@smU', $pattern_cont, $m)) {

		$loc_good = false;

		echo "[ !! ] Does not have a notice beginning with \"Disallocated by IANA\": $tld\n";

	}

}

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

}

# ---

echo "* Check if every TLD has a single whois entry\n";

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

$loc_good = true;

foreach ($tld_cont as $tld) {

	$tld = trim($tld);

	if ($tld == '') continue;

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

	$tld = strtolower($tld);

	$regex = '@[\r\n]\s*(?<!#)\s*[^a-zA-Z\r\n#]*\\\\.'.trim($tld).'\\$\s*[\r\n]@isU';

	$pc = $pattern_cont;

	$pc = preg_replace('@\\(\\?\\:[a-zA-Z\\|]+\\)@isU', '', $pc); # z.B. \.((?:ac|co|net|or|web|sch|mil|go|my|biz)\.id)$

	$pc = str_replace('(', '', $pc);

	$pc = str_replace(')', '', $pc); # wegen regexes, z.b. \.(pn)$

	# if (strpos($pattern_cont, '\.'.trim($tld)) === false) {

	# if (strpos(str_replace(')', '', $pattern_cont), '\.'.trim($tld).'$') === false) { # ')' wegen RegEx'es

	if (!preg_match($regex, $pc)) {

		$iana_url = sprintf(IANA_TLD_ENTRY, $tld);

		$iana_cont = file_get_contents($iana_url);

		if (strpos($iana_cont, 'This domain is not present in the root zone at this time.') === false) {

			$loc_good = false;

			# TODO: hier sollte man evtl schon gucken, ob bei iana.org ein whois-name vorliegt.

			echo "[ !! ] Not in pattern-file: $tld (see $iana_url )\n";

		}

	}

}

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

}

# ---

echo "* Check if TLDs have the latest Root-Whois-Server URLs ( ".ROOT_WHOIS_SERVER_LIST." )\n";

# TODO: geht nicht wenn 2 TLDs sich einen :whois teilen

# geht auch nicht mit domains mit third level tlds oder regexes usw

# TODO: allerdings wird keine kontrolle gemacht ob die auskommentierte Zeile wirklich zu $tld gehört!

echo "Note: For following TLDs, the root whois server may be commented out: ".implode(', ', $anormale_whois)."\n";

$loc_good = true;

$rws_cont = rws_get_list();

foreach ($rws_cont as $tld => $whois) {

	$c = "$tld => $whois";

	$g = false;

	if (in_arrayi($tld, $anormale_whois)) {

		# Sonderfälle... whois server die nicht gehen...

		# Der Server muss dann nur irgendwo vorkommen, in Form einer auskommentierten Zeile

		$g = (

			(stripos($pattern_cont, ':whois|'.$whois."\n") !== false) ||

			(stripos($pattern_cont, 'whois::'.$whois.":::") !== false) || // multiple

			(stripos($pattern_cont, 'whois::'.$whois."\n") !== false) || // multiple

			(stripos($pattern_cont, ':inicwhois|'.$whois."\n") !== false) ||

			(stripos($pattern_cont, 'inicwhois::'.$whois.":::") !== false) || // multiple

			(stripos($pattern_cont, 'inicwhois::'.$whois."\n") !== false) // multiple

		);

	} else {

		# TODO: auch auskommentierte einträge erlauben, falls die RWI outdated ist

		$regex = "@\\n:(inic){0,1}whois\\|".preg_quote($whois, '@')."(\\|.*)*\\n(#.*\n)*\\\\.".preg_quote($tld, '@')."\\$@ismU";

		$g = preg_match($regex, $pattern_cont);

		if (!$g) {

			# Multiple?

			# TODO: bessere pruefung auf auskommentiere eintraege

			$regex = "@(:multiple\\|\\||:::)(inic){0,1}whois::".preg_quote($whois, '@')."(:[^\n]*|)\s*\n(#.*\n)*(\\(\\.\\*\\)){0,1}\\\\.".preg_quote($tld, '@')."\\$@ismU";

			$g = preg_match($regex, $pattern_cont);

		}

	}

	if (!$g) {

		# If the data is not live, we need to check if the whois server is still official (and not changed since then)

		if ((!rws_is_live()) && (iana_get_whois($tld) != $whois)) continue;

		# If we reported it as dead, we trust that. (In case it is wrong, the whois-ping will warn about it)

		if (reported_dead($whois)) continue;

		# Whois servers which are official but are down, are not interesting. Do not warn.

		if (!gwitc_is_port_open($whois, 43)) continue;

		$loc_good = false;

		echo "[ !! ] Not in TLD file: $c\n";

	}

	unset($g);

}

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

}

# ---

echo "* Check if there is a newer official version of gwhois which needs to be merged with the ViaThinkSoft fork ( ".DEBIAN_GWHOIS_PACKAGE_URL." )\n";

define('CUR_VER', gwi_getInstalledBaseVersion());

$latest_official = getLatestGWIversion();

$latest_official = str_replace('-', '.', $latest_official); // they use the reserved '-' character in their versions (used for delimiting upstream versions)

$loc_good = true;

if (CUR_VER != $latest_official) {

	echo '[ !! ] Latest version: '.$latest_official." (This version: ".CUR_VER.")\n";

	$loc_good = false;

}

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

}

# ---

echo "* Check if IANA has (probably) a TLD whois list in ".IANA_TLD_FOLDER."\n";

$x = cached_file(IANA_TLD_FOLDER, CACHE_FILE_DIR);

$loc_good = substr_count($x, '.txt">') == 1;

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

	echo "[ !! ] Something has changed ! Please change rws_*() in the QA tool!\n";

}

# ---

echo "* Check if testcases require attention\n";

$out = array();

exec(__DIR__ . '/testcases/status_short', $out, $code);

$loc_good = $code == 0;

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

	echo trim(implode("\n", $out))."\n";

}

# ---

echo "* Check the status of whois-ping\n";

$out = array();

exec(__DIR__ . '/whois-ping/showerrors', $out, $code);

$loc_good = $code == 0;

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

	echo trim(implode("\n", $out))."\n";

}

# ---

echo "* Check syntax of all script files (Perl, PHP)\n";

$out = array();

exec(__DIR__ . '/syntax/global-syntax-check', $out, $code);

$loc_good = $code == 0;

if ($loc_good) {

	echo "[ OK ]\n";

} else {

	$global_status=1;

	echo trim(implode("\n", $out))."\n";

}

# ---

exit($global_status);

# ---

function getLatestGWIversion() {

	$x = cached_file(DEBIAN_GWHOIS_PACKAGE_URL, CACHE_FILE_DIR);

	preg_match_all('@<br>(.*): all@isU', $x, $m);

	$m = $m[1];

	rsort($m);

	return $m[0]; // biggest version

}

// RWS = Root Whois Server List functions

// (needs to be updated as soon as IANA implements this service.)

function rws_get_list() {

	$result = array();

	$tld_cont = explode("\n", cached_file(ROOT_WHOIS_SERVER_LIST, CACHE_FILE_DIR));

	foreach ($tld_cont as &$c) {

		$c = trim($c);

		if ($c == '') continue;

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

		$ary = explode(" => ", $c);

		$tld = $ary[0];

		$whois = trim($ary[1]);

		$result[$tld] = $whois;

	}

	return $result;

}

function rws_is_live() {

	// The data is collected every week.

	return false;

}

function iana_get_whois($tld) {

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

	if (strpos($out, 'domain:') === false) return false;

	if (!preg_match("@\nwhois:\\s*([^\\s]+)\n@ismU", $out, $m)) return false;

	return $m[1];

}

function gwi_getInstalledBaseVersion() {

	# TODO: auch andere dateien anschauen

	$pattern = file(DIR . '/../../main/pattern/001_subdomains');

	if (!preg_match('@#: version (.+)\+@iU', $pattern[0], $m)) return false;

	return $m[1];

}

function in_arrayi($needle, $haystack) {

	return in_array(strtolower($needle), array_map('strtolower', $haystack));

}

function reported_dead($whois) {

	$dead_servers_raw = file(DIR . '/../config/dead-servers.list');

	foreach ($dead_servers_raw as &$server) {

		$server = trim($server);

		if ($server == '') continue;

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

		if ($server == $whois) return true;

	}

	return false;

}