/**
* Gets the list of included resources as a list of absolute or relative paths of
* resources included in the current item. This allows for a better SCORM export.
* The list will generally include pictures, flash objects, java applets, or any other
* stuff included in the source of the current item. The current item is expected
* to be an HTML file. If it is not, then the function will return and empty list.
* @param string type (one of the Dokeos tools) - optional (otherwise takes the current item's type)
* @param string path (absolute file path) - optional (otherwise takes the current item's path)
* @param int level of recursivity we're in
* @return array List of file paths. An additional field containing 'local' or 'remote' helps determine if the file should be copied into the zip or just linked
*/
function get_resources_from_source($type = null, $abs_path = null, $recursivity = 1)
{
$max = 5;
if ($recursivity > $max) {
return array();
}
if (!isset($type)) {
$type = $this->get_type();
}
if (!isset($abs_path)) {
$path = $this->get_file_path();
$abs_path = api_get_path(SYS_COURSE_PATH) . api_get_course_path() . '/' . $path;
//echo "Abs path coming from item : ".$abs_path."<br/>\n";
}
/*
else
{
echo "Abs path coming from param: ".$abs_path."<br/>\n";
}
*/
//error_log(str_repeat(' ',$recursivity).'Analyse file '.$abs_path,0);
$files_list = array();
$type = $this->get_type();
switch ($type) {
case TOOL_DOCUMENT:
case TOOL_QUIZ:
case 'sco':
//get the document and, if HTML, open it
if (is_file($abs_path)) {
//for now, read the whole file in one go (that's gonna be a problem when the file is too big)
$info = pathinfo($abs_path);
$ext = $info['extension'];
switch (strtolower($ext)) {
case 'html':
case 'htm':
case 'shtml':
case 'css':
$wanted_attributes = array('src', 'url', '@import', 'href', 'value');
//parse it for included resources
$file_content = file_get_contents($abs_path);
//get an array of attributes from the HTML source
$attributes = learnpathItem::parse_HTML_attributes($file_content, $wanted_attributes);
//look at 'src' attributes in this file
foreach ($wanted_attributes as $attr) {
if (isset($attributes[$attr])) {
//find which kind of path these are (local or remote)
$sources = $attributes[$attr];
foreach ($sources as $source) {
//skip what is obviously not a resource
if (strpos($source, "+this.")) {
continue;
}
//javascript code - will still work unaltered
if (strpos($source, '.') === false) {
continue;
}
//no dot, should not be an external file anyway
if (strpos($source, 'mailto:')) {
continue;
}
//mailto link
if (strpos($source, ';') && !strpos($source, '&')) {
continue;
}
//avoid code - that should help
if ($attr == 'value') {
if (strpos($source, 'mp3file')) {
$files_list[] = array(substr($source, 0, strpos($source, '.swf') + 4), 'local', 'abs');
$mp3file = substr($source, strpos($source, 'mp3file=') + 8);
if (substr($mp3file, 0, 1) == '/') {
$files_list[] = array($mp3file, 'local', 'abs');
} else {
$files_list[] = array($mp3file, 'local', 'rel');
}
} elseif (strpos($source, 'flv=') === 0) {
$source = substr($source, 4);
if (strpos($source, '&') > 0) {
$source = substr($source, 0, strpos($source, '&'));
}
if (strpos($source, '://') > 0) {
if (strpos($source, api_get_path(WEB_PATH)) !== false) {
//we found the current portal url
$files_list[] = array($source, 'local', 'url');
} else {
//we didn't find any trace of current portal
$files_list[] = array($source, 'remote', 'url');
}
} else {
$files_list[] = array($source, 'local', 'abs');
}
continue;
//skipping anything else to avoid two entries (while the others can have sub-files in their url, flv's can't)
}
}
if (strpos($source, '://') > 0) {
//cut at '?' in a URL with params
if (strpos($source, '?') > 0) {
$second_part = substr($source, strpos($source, '?'));
if (strpos($second_part, '://') > 0) {
//if the second part of the url contains a url too, treat the second one before cutting
$pos1 = strpos($second_part, '=');
$pos2 = strpos($second_part, '&');
$second_part = substr($second_part, $pos1 + 1, $pos2 - ($pos1 + 1));
if (strpos($second_part, api_get_path(WEB_PATH)) !== false) {
//we found the current portal url
$files_list[] = array($second_part, 'local', 'url');
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $second_part, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} else {
//we didn't find any trace of current portal
$files_list[] = array($second_part, 'remote', 'url');
}
} elseif (strpos($second_part, '=') > 0) {
if (substr($second_part, 0, 1) === '/') {
//link starts with a /, making it absolute (relative to DocumentRoot)
$files_list[] = array($second_part, 'local', 'abs');
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $second_part, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} elseif (strstr($second_part, '..') === 0) {
//link is relative but going back in the hierarchy
$files_list[] = array($second_part, 'local', 'rel');
$dir = dirname($abs_path);
$new_abs_path = realpath($dir . '/' . $second_part);
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $new_abs_path, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} else {
//no starting '/', making it relative to current document's path
if (substr($second_part, 0, 2) == './') {
$second_part = substr($second_part, 2);
}
$files_list[] = array($second_part, 'local', 'rel');
$dir = dirname($abs_path);
$new_abs_path = realpath($dir . '/' . $second_part);
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $new_abs_path, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
}
}
//leave that second part behind now
$source = substr($source, 0, strpos($source, '?'));
if (strpos($source, '://') > 0) {
if (strpos($source, api_get_path(WEB_PATH)) !== false) {
//we found the current portal url
$files_list[] = array($source, 'local', 'url');
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $source, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} else {
//we didn't find any trace of current portal
$files_list[] = array($source, 'remote', 'url');
}
} else {
//no protocol found, make link local
if (substr($source, 0, 1) === '/') {
//link starts with a /, making it absolute (relative to DocumentRoot)
$files_list[] = array($source, 'local', 'abs');
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $source, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} elseif (strstr($source, '..') === 0) {
//link is relative but going back in the hierarchy
$files_list[] = array($source, 'local', 'rel');
$dir = dirname($abs_path);
$new_abs_path = realpath($dir . '/' . $source);
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $new_abs_path, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} else {
//no starting '/', making it relative to current document's path
if (substr($source, 0, 2) == './') {
$source = substr($source, 2);
}
$files_list[] = array($source, 'local', 'rel');
$dir = dirname($abs_path);
$new_abs_path = realpath($dir . '/' . $source);
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $new_abs_path, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
}
}
}
//found some protocol there
if (strpos($source, api_get_path(WEB_PATH)) !== false || strpos(str_replace('http', 'https', $source), api_get_path(WEB_PATH)) !== false) {
//we found the current portal url
// if url of doc is with http:// and is the same as portal but with https, replace source url
if (strpos(str_replace('http', 'https', $source), api_get_path(WEB_PATH)) !== false) {
$source = str_replace('http', 'https', $source);
}
$files_list[] = array($source, 'local', 'url');
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $source, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} else {
//we didn't find any trace of current portal
$files_list[] = array($source, 'remote', 'url');
}
} else {
//no protocol found, make link local
if (substr($source, 0, 1) === '/') {
//link starts with a /, making it absolute (relative to DocumentRoot)
$files_list[] = array($source, 'local', 'abs');
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $source, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} elseif (strstr($source, '..') === 0) {
//link is relative but going back in the hierarchy
$files_list[] = array($source, 'local', 'rel');
$dir = dirname($abs_path);
$new_abs_path = realpath($dir . '/' . $source);
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $new_abs_path, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
} else {
//no starting '/', making it relative to current document's path
if (substr($source, 0, 2) == './') {
$source = substr($source, 2);
}
$files_list[] = array($source, 'local', 'rel');
$dir = dirname($abs_path);
$new_abs_path = realpath($dir . '/' . $source);
$in_files_list[] = learnpathItem::get_resources_from_source(TOOL_DOCUMENT, $new_abs_path, $recursivity + 1);
if (count($in_files_list) > 0) {
$files_list = array_merge($files_list, $in_files_list);
}
}
}
}
}
}
break;
default:
break;
}
} else {
//the file could not be found
return false;
}
break;
default:
//ignore
break;
}
//error_log(str_repeat(' ',$recursivity),'found files '.print_r($files_list,true),0);
//return $files_list;
$checked_files_list = array();
$checked_array_list = array();
foreach ($files_list as $idx => $file) {
if (!empty($file[0])) {
if (!in_array($file[0], $checked_files_list)) {
$checked_files_list[] = $files_list[$idx][0];
$checked_array_list[] = $files_list[$idx];
}
}
}
return $checked_array_list;
}
