lib/htmlpurifier/HTMLPurifier/AttrDef/URI/Host.php

   1 <?php
   2
   3 require_once 'HTMLPurifier/AttrDef.php';
   4 require_once 'HTMLPurifier/AttrDef/URI/IPv4.php';
   5 require_once 'HTMLPurifier/AttrDef/URI/IPv6.php';
   6
   7 /**
   8  * Validates a host according to the IPv4, IPv6 and DNS (future) specifications.
   9  */
  10 class HTMLPurifier_AttrDef_URI_Host extends HTMLPurifier_AttrDef
  11 {
  12
  13     /**
  14      * Instance of HTMLPurifier_AttrDef_URI_IPv4 sub-validator
  15      */
  16     var $ipv4;
  17
  18     /**
  19      * Instance of HTMLPurifier_AttrDef_URI_IPv6 sub-validator
  20      */
  21     var $ipv6;
  22
  23     function HTMLPurifier_AttrDef_URI_Host() {
  24         $this->ipv4 = new HTMLPurifier_AttrDef_URI_IPv4();
  25         $this->ipv6 = new HTMLPurifier_AttrDef_URI_IPv6();
  26     }
  27
  28     function validate($string, $config, &$context) {
  29         $length = strlen($string);
  30         if ($string === '') return '';
  31         if ($length > 1 && $string[0] === '[' && $string[$length-1] === ']') {
  32             //IPv6
  33             $ip = substr($string, 1, $length - 2);
  34             $valid = $this->ipv6->validate($ip, $config, $context);
  35             if ($valid === false) return false;
  36             return '['. $valid . ']';
  37         }
  38
  39         // need to do checks on unusual encodings too
  40         $ipv4 = $this->ipv4->validate($string, $config, $context);
  41         if ($ipv4 !== false) return $ipv4;
  42
  43         // A regular domain name.
  44
  45         // This breaks I18N domain names, but we don't have proper IRI support,
  46         // so force users to insert Punycode. If there's complaining we'll
  47         // try to fix things into an international friendly form.
  48
  49         // The productions describing this are:
  50         $a   = '[a-z]';     // alpha
  51         $an  = '[a-z0-9]';  // alphanum
  52         $and = '[a-z0-9-]'; // alphanum | "-"
  53         // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
  54         $domainlabel   = "$an($and*$an)?";
  55         // toplabel    = alpha | alpha *( alphanum | "-" ) alphanum
  56         $toplabel      = "$a($and*$an)?";
  57         // hostname    = *( domainlabel "." ) toplabel [ "." ]
  58         $match = preg_match("/^($domainlabel\.)*$toplabel\.?$/i", $string);
  59         if (!$match) return false;
  60
  61         return $string;
  62     }
  63
  64 }
  65