library/HTMLPurifier/URIFilter/MakeAbsolute.php

   1 <?php
   2
   3 // does not support network paths
   4
   5 require_once 'HTMLPurifier/URIFilter.php';
   6
   7 HTMLPurifier_ConfigSchema::define(
   8     'URI', 'MakeAbsolute', false, 'bool', '
   9 <p>
  10     Converts all URIs into absolute forms. This is useful when the HTML
  11     being filtered assumes a specific base path, but will actually be
  12     viewed in a different context (and setting an alternate base URI is
  13     not possible). %URI.Base must be set for this directive to work.
  14     This directive has been available since 2.1.0.
  15 </p>
  16 ');
  17
  18 class HTMLPurifier_URIFilter_MakeAbsolute extends HTMLPurifier_URIFilter
  19 {
  20     public $name = 'MakeAbsolute';
  21     protected $base;
  22     protected $basePathStack = array();
  23     public function prepare($config) {
  24         $def = $config->getDefinition('URI');
  25         $this->base = $def->base;
  26         if (is_null($this->base)) {
  27             trigger_error('URI.MakeAbsolute is being ignored due to lack of value for URI.Base configuration', E_USER_ERROR);
  28             return;
  29         }
  30         $this->base->fragment = null; // fragment is invalid for base URI
  31         $stack = explode('/', $this->base->path);
  32         array_pop($stack); // discard last segment
  33         $stack = $this->_collapseStack($stack); // do pre-parsing
  34         $this->basePathStack = $stack;
  35     }
  36     public function filter(&$uri, $config, $context) {
  37         if (is_null($this->base)) return true; // abort early
  38         if (
  39             $uri->path === '' && is_null($uri->scheme) &&
  40             is_null($uri->host) && is_null($uri->query) && is_null($uri->fragment)
  41         ) {
  42             // reference to current document
  43             $uri = $this->base->copy();
  44             return true;
  45         }
  46         if (!is_null($uri->scheme)) {
  47             // absolute URI already: don't change
  48             if (!is_null($uri->host)) return true;
  49             $scheme_obj = $uri->getSchemeObj($config, $context);
  50             if (!$scheme_obj) {
  51                 // scheme not recognized
  52                 return false;
  53             }
  54             if (!$scheme_obj->hierarchical) {
  55                 // non-hierarchal URI with explicit scheme, don't change
  56                 return true;
  57             }
  58             // special case: had a scheme but always is hierarchical and had no authority
  59         }
  60         if (!is_null($uri->host)) {
  61             // network path, don't bother
  62             return true;
  63         }
  64         if ($uri->path === '') {
  65             $uri->path = $this->base->path;
  66         }elseif ($uri->path[0] !== '/') {
  67             // relative path, needs more complicated processing
  68             $stack = explode('/', $uri->path);
  69             $new_stack = array_merge($this->basePathStack, $stack);
  70             $new_stack = $this->_collapseStack($new_stack);
  71             $uri->path = implode('/', $new_stack);
  72         }
  73         // re-combine
  74         $uri->scheme = $this->base->scheme;
  75         if (is_null($uri->userinfo)) $uri->userinfo = $this->base->userinfo;
  76         if (is_null($uri->host))     $uri->host     = $this->base->host;
  77         if (is_null($uri->port))     $uri->port     = $this->base->port;
  78         return true;
  79     }
  80
  81     /**
  82      * Resolve dots and double-dots in a path stack
  83      */
  84     private function _collapseStack($stack) {
  85         $result = array();
  86         for ($i = 0; isset($stack[$i]); $i++) {
  87             $is_folder = false;
  88             // absorb an internally duplicated slash
  89             if ($stack[$i] == '' && $i && isset($stack[$i+1])) continue;
  90             if ($stack[$i] == '..') {
  91                 if (!empty($result)) {
  92                     $segment = array_pop($result);
  93                     if ($segment === '' && empty($result)) {
  94                         // error case: attempted to back out too far:
  95                         // restore the leading slash
  96                         $result[] = '';
  97                     } elseif ($segment === '..') {
  98                         $result[] = '..'; // cannot remove .. with ..
  99                     }
 100                 } else {
 101                     // relative path, preserve the double-dots
 102                     $result[] = '..';
 103                 }
 104                 $is_folder = true;
 105                 continue;
 106             }
 107             if ($stack[$i] == '.') {
 108                 // silently absorb
 109                 $is_folder = true;
 110                 continue;
 111             }
 112             $result[] = $stack[$i];
 113         }
 114         if ($is_folder) $result[] = '';
 115         return $result;
 116     }
 117 }
 118