include/classes/ContentOutputParser.class.php

   1 <?php
   2 /************************************************************************/
   3 /* ATutor                                                               */
   4 /************************************************************************/
   5 /* Copyright (c) 2002 - 2009                                            */
   6 /* Inclusive Design Institute                                           */
   7 /*                                                                      */
   8 /* This program is free software. You can redistribute it and/or        */
   9 /* modify it under the terms of the GNU General Public License          */
  10 /* as published by the Free Software Foundation.                        */
  11 /************************************************************************/
  12 // $Id: ContentOutputParser.class.php 10278 2010-10-04 16:47:45Z hwong $
  13
  14 class ContentOutputParser {
  15     function ContentOutputParser(){}
  16     function openHandler(& $parser,$name,$attrs) {
  17                 global $my_files;
  18
  19                 $name = strtolower($name);
  20                 $attrs = array_change_key_case($attrs, CASE_LOWER);
  21
  22         /*
  23             the following resources are to be identified:
  24             even if some of these can't be images, they can still be files in the content dir.
  25             theoretically the only urls we wouldn't deal with would be for a <!DOCTYPE and <form>
  26
  27             img         => src
  28             a           => href                         // ignore if href doesn't exist (ie. <a name>)
  29             object      => data | classid       // probably only want data
  30             applet      => classid | archive                    // whatever these two are should double check to see if it's a valid file (not a dir)
  31             link        => href
  32             script      => src
  33             form        => action
  34             input       => src
  35             iframe      => src
  36
  37         */
  38                 $elements = array(      'img'           => 'src',
  39                                                         'a'                     => 'href',
  40                                                         'object'        => array('data', 'classid'),
  41                                                         'applet'        => array('classid', 'archive'),
  42                                                         'link'          => 'href',
  43                                                         'script'        => 'src',
  44                                                         'form'          => 'action',
  45                                                         'input'         => 'src',
  46                                                         'iframe'        => 'src',
  47                                                         'embed'         => 'src',
  48                                                         'param'         => 'value');
  49
  50                 /* check if this attribute specifies the files in different ways: (ie. java) */
  51                 if (is_array($elements[$name])) {
  52                         $items = $elements[$name];
  53                         foreach ($items as $item) {
  54                                 if ($attrs[$item] != '') {
  55
  56                                         /* some attributes allow a listing of files to include seperated by commas (ie. applet->archive). */
  57                                         if (strpos($attrs[$item], ',') !== false) {
  58                                                 $files = explode(',', $attrs[$item]);
  59                                                 foreach ($files as $file) {
  60                                                         $my_files[] = trim($file);
  61                                                 }
  62                                         } else {
  63                                                 $file = trim($attrs[$item]);
  64                                                 // filter our classid="clsid:..."
  65                                                 if (!strpos($file, "clsid:")) {
  66                                                         $my_files[] = $file;
  67                                                 }
  68                                         }
  69                                 }
  70                         }
  71                 } else if (isset($elements[$name]) && ($attrs[$elements[$name]] != '')) {
  72                         //hack, if param[name]=src or none <param> tag, extract. Skip all other <param> attributes.
  73                         if ($name!='param' || $attrs['name']=='src'){
  74                                 //skip glossary.html, tweak to accomodate atutor imscp; also skip repeated entries.
  75                                 //skip javascript: links, void();, #, mailto:
  76                                 if (strpos($attrs[$elements[$name]], 'glossary.html')===false
  77                                     && !in_array($attrs[$elements[$name]], $my_files)
  78                                     && $attrs[$elements[$name]]!='#'
  79                                     && strpos($attrs[$elements[$name]], 'javascript:')===false
  80                                     && strpos($attrs[$elements[$name]], 'mailto:')===false
  81                                     && strpos($attrs[$elements[$name]], 'void(')===false
  82                                    ){
  83                                         $my_files[] = $attrs[$elements[$name]];
  84                                 }
  85                         }
  86                 }
  87         }
  88
  89         function closeHandler(& $parser,$name) { }
  90 }
  91 ?>