lib/htmlpurifier/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 breaks I18N domain names, but we don't have proper IRI support,
  48         // so force users to insert Punycode. If there's complaining we'll
  49         // try to fix things into an international friendly form.
  50
  51         // The productions describing this are:
  52         $a   = '[a-z]';     // alpha
  53         $an  = '[a-z0-9]';  // alphanum
  54         $and = '[a-z0-9-]'; // alphanum | "-"
  55         // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
  56         $domainlabel   = "$an($and*$an)?";
  57         // toplabel    = alpha | alpha *( alphanum | "-" ) alphanum
  58         $toplabel      = "$a($and*$an)?";
  59         // hostname    = *( domainlabel "." ) toplabel [ "." ]
  60         $match = preg_match("/^($domainlabel\.)*$toplabel\.?$/i", $string);
  61         if (!$match) return false;
  62
  63         return $string;
  64     }
  65
  66 }
  67
  68 // vim: et sw=4 sts=4