docs/include/lib/html_resource_parser.inc.php

   1 <?php
   2 /****************************************************************/
   3 /* ATutor                                                                                                               */
   4 /****************************************************************/
   5 /* Copyright (c) 2002-2008 by Greg Gay, Cindy Qi Li,                    */
   6 /* & Harris Wong                                                                                                */
   7 /* Inclusive Design Institute                                   */
   8 /* http://atutor.ca                                                                                             */
   9 /*                                                              */
  10 /* This program is free software. You can redistribute it and/or*/
  11 /* modify it under the terms of the GNU General Public License  */
  12 /* as published by the Free Software Foundation.                                */
  13 /****************************************************************/
  14 // $Id$
  15
  16 /**
  17  * @param
  18  * @param   int     course id, which should be in $_SESSION, but during export to AContent,
  19  *                  SESSION is not available, thus we will have to use parameters
  20  */
  21 function get_html_resources($text, $course_id=0) {
  22         $resources = array();
  23     if ($course_id == 0){
  24         $course_id = $_SESSION['course_id'];
  25     }
  26
  27         $handler = new XML_HTMLSax_Handler();
  28
  29         $parser = new XML_HTMLSax();
  30         $parser->set_object($handler);
  31         $parser->set_element_handler('openHandler','closeHandler');
  32
  33         $parser->parse($text);
  34
  35         foreach ($handler->resources as $resource) {
  36                 $url_parts = @parse_url($resource);
  37
  38                 if (isset($url_parts['scheme'])) {
  39                         // we don't want full urls
  40                         continue;
  41                 }
  42
  43                 if ((substr($resource, 0, 1) == '/')) {
  44                         // we don't want absolute urls
  45                         continue;
  46                 }
  47
  48                 // make sure this resource exists in this course's content directory:
  49                 $resource_server_path = realpath(AT_CONTENT_DIR . $course_id. '/' . $resource);
  50                 if (file_exists($resource_server_path) && is_file($resource_server_path)) {
  51                         $resources[$resource] = $resource_server_path;
  52                 }
  53         }
  54         return $resources;
  55 }
  56
  57 /*
  58         the following resources are to be identified:
  59         even if some of these can't be images, they can still be files in the content dir.
  60         theoretically the only urls we wouldn't deal with would be for a <!DOCTYPE and <form>
  61
  62         img             => src
  63         a               => href                         // ignore if href doesn't exist (ie. <a name>)
  64         object  => data | classid       // probably only want data
  65         applet  => classid | archive                    // whatever these two are should double check to see if it's a valid file (not a dir)
  66         link    => href
  67         script  => src
  68         form    => action
  69         input   => src
  70         iframe  => src
  71 */
  72 class XML_HTMLSax_Handler {
  73         var $elements = array(  'img'    => 'src',
  74                                                         'a'              => 'href',
  75                                                         'object' =>  array('data',    'classid'),
  76                                                         'applet' =>  array('classid', 'archive'),
  77                                                         'link'   => 'href',
  78                                                         'script' => 'src',
  79                                                         'form'   => 'action',
  80                                                         'input'  => 'src',
  81                                                         'iframe' => 'src',
  82                                                         'embed'  => 'src',
  83                                                         'param'  => 'value');
  84         var $resources = array();
  85
  86     function XML_HTMLSax_Handler() {
  87                 $this->resources = array();
  88         }
  89
  90     function openHandler(& $parser,$name,$attrs) {
  91                 $name = strtolower($name);
  92                 $attrs = array_change_key_case($attrs, CASE_LOWER);
  93
  94                 /* check if this attribute specifies the files in different ways: (ie. java) */
  95                 if (is_array($this->elements[$name])) {
  96                         $items = $this->elements[$name];
  97
  98                         foreach ($items as $item) {
  99                                 if ($attrs[$item] != '') {
 100                                         /* some attributes allow a listing of files to include seperated by commas (ie. applet->archive). */
 101                                         if (strpos($attrs[$item], ',') !== false) {
 102                                                 $files = explode(',', $attrs[$item]);
 103                                                 foreach ($files as $file) {
 104                                                         $this->resources[] = trim($file);
 105                                                 }
 106                                         } else {
 107                                                 $this->resources[] = $attrs[$item];
 108                                         }
 109                                 }
 110                         }
 111                 } else if (isset($this->elements[$name]) && ($attrs[$this->elements[$name]] != '')) {
 112                         /* we know exactly which attribute contains the reference to the file. */
 113                         $this->resources[] = $attrs[$this->elements[$name]];
 114                 }
 115     }
 116     function closeHandler(& $parser,$name) { }
 117 }
 118 ?>