library/HTMLPurifier/URIFilter/MakeAbsolute.php

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