mods/_core/imsafa/html/resources_parser.inc.php

   1 <?php
   2 /****************************************************************/
   3 /* ATutor                                                                                                               */
   4 /****************************************************************/
   5 /* Copyright (c) 2002-2010                                      */
   6 /* Inclusive Design Institute                                   */
   7 /* http://atutor.ca                                                                                             */
   8 /*                                                              */
   9 /* This program is free software. You can redistribute it and/or*/
  10 /* modify it under the terms of the GNU General Public License  */
  11 /* as published by the Free Software Foundation.                                */
  12 /****************************************************************/
  13 // $Id: resources_parser_inc.inc.php 7208 2008-07-04 16:07:24Z silvia $
  14
  15 if (!defined('AT_INCLUDE_PATH')) { exit; }
  16 //require(AT_INCLUDE_PATH.'lib/output.inc.php');
  17
  18 global $db;
  19
  20 define('AT_INCLUDE_PATH', '../include/');
  21
  22 //echo 'uffa';
  23 //echo $_POST['body_text'];
  24
  25 $body_text      = htmlspecialchars($stripslashes($_POST['body_text']));
  26 $body_t         = html_entity_decode($body_text);
  27
  28
  29 require(AT_INCLUDE_PATH.'classes/XML/XML_HTMLSax/XML_HTMLSax.php');     /* for XML_HTMLSax */
  30 require(AT_INCLUDE_PATH.'../mods/_core/imscp/include/ims_template.inc.php');                            /* for ims templates + print_organizations() */
  31
  32 /*
  33 the following resources are to be identified:
  34 even if some of these can't be images, they can still be files in the content dir.
  35 theoretically the only urls we wouldn't deal with would be for a <!DOCTYPE and <form>
  36
  37 img             => src
  38 a               => href                         // ignore if href doesn't exist (ie. <a name>)
  39 object  => data | classid       // probably only want data
  40 applet  => classid | archive                    // whatever these two are should double check to see if it's a valid file (not a dir)
  41 script  => src
  42 input   => src
  43 iframe  => src
  44 */
  45
  46 class MyHandler {
  47         function MyHandler(){}
  48         function openHandler(& $parser,$name,$attrs) {
  49                 global $my_files;
  50
  51                 $name = strtolower($name);
  52                 $attrs = array_change_key_case($attrs, CASE_LOWER);
  53
  54                 $elements = array(      'img'           => 'src',
  55                                                         'a'                     => 'href',
  56                                                         'object'        => array('data', 'classid'),
  57                                                         'applet'        => array('classid', 'archive'),
  58                                                         'script'        => 'src',
  59                                                         'input'         => 'src',
  60                                                         'iframe'        => 'src',
  61                                                         'embed'         => 'src',
  62                                                         );
  63
  64                 /* check if this attribute specifies the files in different ways: (ie. java) */
  65                 if (is_array($elements[$name])) {
  66                         $items = $elements[$name];
  67
  68                         foreach ($items as $item) {
  69                                 if ($attrs[$item] != '') {
  70
  71                                         /* some attributes allow a listing of files to include seperated by commas (ie. applet->archive). */
  72                                         if (strpos($attrs[$item], ',') !== false) {
  73                                                 $files = explode(',', $attrs[$item]);
  74                                                 foreach ($files as $file) {
  75                                                         $my_files[] = trim($file);
  76                                                 }
  77                                         } else {
  78                                                 $my_files[] = $attrs[$item];
  79                                         }
  80                                 }
  81                         }
  82                 } else if (isset($elements[$name]) && ($attrs[$elements[$name]] != '')) {
  83                         /* we know exactly which attribute contains the reference to the file. */
  84                         $my_files[] = $attrs[$elements[$name]];
  85                 }
  86         }
  87         function closeHandler(& $parser,$name) { }
  88         }
  89
  90 /* get all the content */
  91 $handler=new MyHandler();
  92 $parser = new XML_HTMLSax();
  93 $parser->set_object($handler);
  94 $parser->set_element_handler('openHandler','closeHandler');
  95
  96 /* generate the resources and save the HTML files */
  97
  98 ob_start();
  99
 100 global $parser, $my_files;
 101 global $course_id;
 102
 103 /* add the resource dependancies */
 104 $my_files               = array();
 105 $content_files  = "\n";
 106
 107 //in order to control if some [media] is in the body_text
 108 $body = embed_media($body_t);
 109
 110 $parser->parse($body);
 111
 112 // add by Cindy Li.
 113 // This resolves the problem introduced by [media] tag: when [media] is
 114 // parsed into <object>, same resource appears a few times in <object> with different
 115 // format to cater for different browsers or players. This way creates prolem that different
 116 // formats in <object> are all parsed and considered as different resource. array_unique()
 117 // call solves this problem. But, it introduces the new problem that when a same resource
 118 // appears at different places in the content and users do want to have them with different
 119 // alternatives. With this solution, this same resource only shows up once at "adapt content"
 120 // and only can have one alternative associate with. Table and scripts need to re-design
 121 // to solve this problem, for example, include line number in table.
 122 $my_files = array_unique($my_files);
 123
 124 /* handle @import */
 125 $import_files   = get_import_files($body);
 126
 127 if (count($import_files) > 0) $my_files = array_merge($my_files, $import_files);
 128
 129 $i=0;
 130
 131 foreach ($my_files as $file) {
 132         /* filter out full urls */
 133         $url_parts = @parse_url($file);
 134 //      if (isset($url_parts['scheme']) && substr($file, 0, strlen(AT_BASE_HREF)) != AT_BASE_HREF) {
 135 //              continue;
 136 //      }
 137
 138         /* file should be relative to content. let's double check */
 139         if ((substr($file, 0, 1) == '/')) {
 140                 continue;
 141         }
 142
 143         $resources[$i] = $file;
 144         $i++;
 145 }
 146
 147 $organizations_str = ob_get_contents();
 148 ob_end_clean();