library/HTMLPurifier/URIScheme.php

   1 <?php
   2
   3 /**
   4  * Validator for the components of a URI for a specific scheme
   5  */
   6 abstract class HTMLPurifier_URIScheme
   7 {
   8
   9     /**
  10      * Scheme's default port (integer).  If an explicit port number is
  11      * specified that coincides with the default port, it will be
  12      * elided.
  13      */
  14     public $default_port = null;
  15
  16     /**
  17      * Whether or not URIs of this schem are locatable by a browser
  18      * http and ftp are accessible, while mailto and news are not.
  19      */
  20     public $browsable = false;
  21
  22     /**
  23      * Whether or not the URI always uses <hier_part>, resolves edge cases
  24      * with making relative URIs absolute
  25      */
  26     public $hierarchical = false;
  27
  28     /**
  29      * Whether or not the URI may omit a hostname when the scheme is
  30      * explicitly specified, ala file:///path/to/file. As of writing,
  31      * 'file' is the only scheme that browsers support his properly.
  32      */
  33     public $may_omit_host = false;
  34
  35     /**
  36      * Validates the components of a URI for a specific scheme.
  37      * @param $uri Reference to a HTMLPurifier_URI object
  38      * @param $config HTMLPurifier_Config object
  39      * @param $context HTMLPurifier_Context object
  40      * @return Bool success or failure
  41      */
  42     public abstract function doValidate(&$uri, $config, $context);
  43
  44     /**
  45      * Public interface for validating components of a URI.  Performs a
  46      * bunch of default actions. Don't overload this method.
  47      * @param $uri Reference to a HTMLPurifier_URI object
  48      * @param $config HTMLPurifier_Config object
  49      * @param $context HTMLPurifier_Context object
  50      * @return Bool success or failure
  51      */
  52     public function validate(&$uri, $config, $context) {
  53         if ($this->default_port == $uri->port) $uri->port = null;
  54         // kludge: browsers do funny things when the scheme but not the
  55         // authority is set
  56         if (!$this->may_omit_host &&
  57             // if the scheme is present, a missing host is always in error
  58             (!is_null($uri->scheme) && ($uri->host === '' || is_null($uri->host))) ||
  59             // if the scheme is not present, a *blank* host is in error,
  60             // since this translates into '///path' which most browsers
  61             // interpret as being 'http://path'.
  62              (is_null($uri->scheme) && $uri->host === '')
  63         ) {
  64             do {
  65                 if (is_null($uri->scheme)) {
  66                     if (substr($uri->path, 0, 2) != '//') {
  67                         $uri->host = null;
  68                         break;
  69                     }
  70                     // URI is '////path', so we cannot nullify the
  71                     // host to preserve semantics.  Try expanding the
  72                     // hostname instead (fall through)
  73                 }
  74                 // first see if we can manually insert a hostname
  75                 $host = $config->get('URI.Host');
  76                 if (!is_null($host)) {
  77                     $uri->host = $host;
  78                 } else {
  79                     // we can't do anything sensible, reject the URL.
  80                     return false;
  81                 }
  82             } while (false);
  83         }
  84         return $this->doValidate($uri, $config, $context);
  85     }
  86
  87 }
  88
  89 // vim: et sw=4 sts=4