library/HTMLPurifier/AttrDef/URI/Host.php

   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      * Instance of HTMLPurifier_AttrDef_URI_IPv4 sub-validator
  11      */
  12     protected $ipv4;
  13
  14     /**
  15      * Instance of HTMLPurifier_AttrDef_URI_IPv6 sub-validator
  16      */
  17     protected $ipv6;
  18
  19     public function __construct() {
  20         $this->ipv4 = new HTMLPurifier_AttrDef_URI_IPv4();
  21         $this->ipv6 = new HTMLPurifier_AttrDef_URI_IPv6();
  22     }
  23
  24     public function validate($string, $config, $context) {
  25         $length = strlen($string);
  26         // empty hostname is OK; it's usually semantically equivalent:
  27         // the default host as defined by a URI scheme is used:
  28         //
  29         //      If the URI scheme defines a default for host, then that
  30         //      default applies when the host subcomponent is undefined
  31         //      or when the registered name is empty (zero length).
  32         if ($string === '') return '';
  33         if ($length > 1 && $string[0] === '[' && $string[$length-1] === ']') {
  34             //IPv6
  35             $ip = substr($string, 1, $length - 2);
  36             $valid = $this->ipv6->validate($ip, $config, $context);
  37             if ($valid === false) return false;
  38             return '['. $valid . ']';
  39         }
  40
  41         // need to do checks on unusual encodings too
  42         $ipv4 = $this->ipv4->validate($string, $config, $context);
  43         if ($ipv4 !== false) return $ipv4;
  44
  45         // A regular domain name.
  46
  47         // This doesn't match I18N domain names, but we don't have proper IRI support,
  48         // so force users to insert Punycode.
  49
  50         // The productions describing this are:
  51         $a   = '[a-z]';     // alpha
  52         $an  = '[a-z0-9]';  // alphanum
  53         $and = '[a-z0-9-]'; // alphanum | "-"
  54         // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
  55         $domainlabel   = "$an($and*$an)?";
  56         // toplabel    = alpha | alpha *( alphanum | "-" ) alphanum
  57         $toplabel      = "$a($and*$an)?";
  58         // hostname    = *( domainlabel "." ) toplabel [ "." ]
  59         if (preg_match("/^($domainlabel\.)*$toplabel\.?$/i", $string)) {
  60             return $string;
  61         }
  62
  63         // If we have Net_IDNA2 support, we can support IRIs by
  64         // punycoding them. (This is the most portable thing to do,
  65         // since otherwise we have to assume browsers support
  66
  67         if ($config->get('Core.EnableIDNA')) {
  68             $idna = new Net_IDNA2(array('encoding' => 'utf8', 'overlong' => false, 'strict' => true));
  69             // we need to encode each period separately
  70             $parts = explode('.', $string);
  71             try {
  72                 $new_parts = array();
  73                 foreach ($parts as $part) {
  74                     $encodable = false;
  75                     for ($i = 0, $c = strlen($part); $i < $c; $i++) {
  76                         if (ord($part[$i]) > 0x7a) {
  77                             $encodable = true;
  78                             break;
  79                         }
  80                     }
  81                     if (!$encodable) {
  82                         $new_parts[] = $part;
  83                     } else {
  84                         $new_parts[] = $idna->encode($part);
  85                     }
  86                 }
  87                 $string = implode('.', $new_parts);
  88                 if (preg_match("/^($domainlabel\.)*$toplabel\.?$/i", $string)) {
  89                     return $string;
  90                 }
  91             } catch (Exception $e) {
  92                 // XXX error reporting
  93             }
  94         }
  95
  96         return false;
  97     }
  98
  99 }
 100
 101 // vim: et sw=4 sts=4