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, $stack);
  58             if ($new_stack[0] !== '' && !is_null($this->base->host)) {
  59                 array_unshift($new_stack, '');
  60             }
  61             $new_stack = $this->_collapseStack($new_stack);
  62             $uri->path = implode('/', $new_stack);
  63         } else {
  64             // absolute path, but still we should collapse
  65             $uri->path = implode('/', $this->_collapseStack(explode('/', $uri->path)));
  66         }
  67         // re-combine
  68         $uri->scheme = $this->base->scheme;
  69         if (is_null($uri->userinfo)) $uri->userinfo = $this->base->userinfo;
  70         if (is_null($uri->host))     $uri->host     = $this->base->host;
  71         if (is_null($uri->port))     $uri->port     = $this->base->port;
  72         return true;
  73     }
  74
  75     /**
  76      * Resolve dots and double-dots in a path stack
  77      */
  78     private function _collapseStack($stack) {
  79         $result = array();
  80         $is_folder = false;
  81         for ($i = 0; isset($stack[$i]); $i++) {
  82             $is_folder = false;
  83             // absorb an internally duplicated slash
  84             if ($stack[$i] == '' && $i && isset($stack[$i+1])) continue;
  85             if ($stack[$i] == '..') {
  86                 if (!empty($result)) {
  87                     $segment = array_pop($result);
  88                     if ($segment === '' && empty($result)) {
  89                         // error case: attempted to back out too far:
  90                         // restore the leading slash
  91                         $result[] = '';
  92                     } elseif ($segment === '..') {
  93                         $result[] = '..'; // cannot remove .. with ..
  94                     }
  95                 } else {
  96                     // relative path, preserve the double-dots
  97                     $result[] = '..';
  98                 }
  99                 $is_folder = true;
 100                 continue;
 101             }
 102             if ($stack[$i] == '.') {
 103                 // silently absorb
 104                 $is_folder = true;
 105                 continue;
 106             }
 107             $result[] = $stack[$i];
 108         }
 109         if ($is_folder) $result[] = '';
 110         return $result;
 111     }
 112 }
 113
 114 // vim: et sw=4 sts=4