| 1 |
efrain |
1 |
<?php
|
|
|
2 |
|
|
|
3 |
/**
|
|
|
4 |
* Validates a host according to the IPv4, IPv6 and DNS (future) specifications.
|
|
|
5 |
*/
|
|
|
6 |
class HTMLPurifier_AttrDef_URI_Host extends HTMLPurifier_AttrDef
|
|
|
7 |
{
|
|
|
8 |
|
|
|
9 |
/**
|
|
|
10 |
* IPv4 sub-validator.
|
|
|
11 |
* @type HTMLPurifier_AttrDef_URI_IPv4
|
|
|
12 |
*/
|
|
|
13 |
protected $ipv4;
|
|
|
14 |
|
|
|
15 |
/**
|
|
|
16 |
* IPv6 sub-validator.
|
|
|
17 |
* @type HTMLPurifier_AttrDef_URI_IPv6
|
|
|
18 |
*/
|
|
|
19 |
protected $ipv6;
|
|
|
20 |
|
|
|
21 |
public function __construct()
|
|
|
22 |
{
|
|
|
23 |
$this->ipv4 = new HTMLPurifier_AttrDef_URI_IPv4();
|
|
|
24 |
$this->ipv6 = new HTMLPurifier_AttrDef_URI_IPv6();
|
|
|
25 |
}
|
|
|
26 |
|
|
|
27 |
/**
|
|
|
28 |
* @param string $string
|
|
|
29 |
* @param HTMLPurifier_Config $config
|
|
|
30 |
* @param HTMLPurifier_Context $context
|
|
|
31 |
* @return bool|string
|
|
|
32 |
*/
|
|
|
33 |
public function validate($string, $config, $context)
|
|
|
34 |
{
|
|
|
35 |
$length = strlen($string);
|
|
|
36 |
// empty hostname is OK; it's usually semantically equivalent:
|
|
|
37 |
// the default host as defined by a URI scheme is used:
|
|
|
38 |
//
|
|
|
39 |
// If the URI scheme defines a default for host, then that
|
|
|
40 |
// default applies when the host subcomponent is undefined
|
|
|
41 |
// or when the registered name is empty (zero length).
|
|
|
42 |
if ($string === '') {
|
|
|
43 |
return '';
|
|
|
44 |
}
|
|
|
45 |
if ($length > 1 && $string[0] === '[' && $string[$length - 1] === ']') {
|
|
|
46 |
//IPv6
|
|
|
47 |
$ip = substr($string, 1, $length - 2);
|
|
|
48 |
$valid = $this->ipv6->validate($ip, $config, $context);
|
|
|
49 |
if ($valid === false) {
|
|
|
50 |
return false;
|
|
|
51 |
}
|
|
|
52 |
return '[' . $valid . ']';
|
|
|
53 |
}
|
|
|
54 |
|
|
|
55 |
// need to do checks on unusual encodings too
|
|
|
56 |
$ipv4 = $this->ipv4->validate($string, $config, $context);
|
|
|
57 |
if ($ipv4 !== false) {
|
|
|
58 |
return $ipv4;
|
|
|
59 |
}
|
|
|
60 |
|
|
|
61 |
// A regular domain name.
|
|
|
62 |
|
|
|
63 |
// This doesn't match I18N domain names, but we don't have proper IRI support,
|
|
|
64 |
// so force users to insert Punycode.
|
|
|
65 |
|
| 1441 |
ariadna |
66 |
// Underscores defined as Unreserved Characters in RFC 3986 are
|
|
|
67 |
// allowed in a URI. There are cases where we want to consider a
|
|
|
68 |
// URI containing "_" such as "_dmarc.example.com".
|
|
|
69 |
// Underscores are not allowed in the default. If you want to
|
|
|
70 |
// allow it, set Core.AllowHostnameUnderscore to true.
|
| 1 |
efrain |
71 |
$underscore = $config->get('Core.AllowHostnameUnderscore') ? '_' : '';
|
|
|
72 |
|
|
|
73 |
// Based off of RFC 1738, but amended so that
|
|
|
74 |
// as per RFC 3696, the top label need only not be all numeric.
|
|
|
75 |
// The productions describing this are:
|
|
|
76 |
$a = '[a-z]'; // alpha
|
| 1441 |
ariadna |
77 |
$an = "[a-z0-9$underscore]"; // alphanum
|
| 1 |
efrain |
78 |
$and = "[a-z0-9-$underscore]"; // alphanum | "-"
|
|
|
79 |
// domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
|
|
|
80 |
$domainlabel = "$an(?:$and*$an)?";
|
|
|
81 |
// AMENDED as per RFC 3696
|
|
|
82 |
// toplabel = alphanum | alphanum *( alphanum | "-" ) alphanum
|
|
|
83 |
// side condition: not all numeric
|
|
|
84 |
$toplabel = "$an(?:$and*$an)?";
|
|
|
85 |
// hostname = *( domainlabel "." ) toplabel [ "." ]
|
|
|
86 |
if (preg_match("/^(?:$domainlabel\.)*($toplabel)\.?$/i", $string, $matches)) {
|
|
|
87 |
if (!ctype_digit($matches[1])) {
|
|
|
88 |
return $string;
|
|
|
89 |
}
|
|
|
90 |
}
|
|
|
91 |
|
|
|
92 |
// PHP 5.3 and later support this functionality natively
|
|
|
93 |
if (function_exists('idn_to_ascii')) {
|
|
|
94 |
if (defined('IDNA_NONTRANSITIONAL_TO_ASCII') && defined('INTL_IDNA_VARIANT_UTS46')) {
|
|
|
95 |
$string = idn_to_ascii($string, IDNA_NONTRANSITIONAL_TO_ASCII, INTL_IDNA_VARIANT_UTS46);
|
|
|
96 |
} else {
|
|
|
97 |
$string = idn_to_ascii($string);
|
|
|
98 |
}
|
|
|
99 |
|
|
|
100 |
// If we have Net_IDNA2 support, we can support IRIs by
|
|
|
101 |
// punycoding them. (This is the most portable thing to do,
|
|
|
102 |
// since otherwise we have to assume browsers support
|
|
|
103 |
} elseif ($config->get('Core.EnableIDNA') && class_exists('Net_IDNA2')) {
|
|
|
104 |
$idna = new Net_IDNA2(array('encoding' => 'utf8', 'overlong' => false, 'strict' => true));
|
|
|
105 |
// we need to encode each period separately
|
|
|
106 |
$parts = explode('.', $string);
|
|
|
107 |
try {
|
|
|
108 |
$new_parts = array();
|
|
|
109 |
foreach ($parts as $part) {
|
|
|
110 |
$encodable = false;
|
|
|
111 |
for ($i = 0, $c = strlen($part); $i < $c; $i++) {
|
|
|
112 |
if (ord($part[$i]) > 0x7a) {
|
|
|
113 |
$encodable = true;
|
|
|
114 |
break;
|
|
|
115 |
}
|
|
|
116 |
}
|
|
|
117 |
if (!$encodable) {
|
|
|
118 |
$new_parts[] = $part;
|
|
|
119 |
} else {
|
|
|
120 |
$new_parts[] = $idna->encode($part);
|
|
|
121 |
}
|
|
|
122 |
}
|
|
|
123 |
$string = implode('.', $new_parts);
|
|
|
124 |
} catch (Exception $e) {
|
|
|
125 |
// XXX error reporting
|
|
|
126 |
}
|
|
|
127 |
}
|
|
|
128 |
// Try again
|
|
|
129 |
if (preg_match("/^($domainlabel\.)*$toplabel\.?$/i", $string)) {
|
|
|
130 |
return $string;
|
|
|
131 |
}
|
|
|
132 |
return false;
|
|
|
133 |
}
|
|
|
134 |
}
|
|
|
135 |
|
|
|
136 |
// vim: et sw=4 sts=4
|