private static function apply(simple_html_dom $dom, $attribute, $elements, ProxyHttpRequest $request)
{
foreach ($dom->find($elements . '[' . $attribute . ']') as $element) {
$attr_val = $element->getAttribute($attribute);
if ($attr_val) {
// Ignore these links.
if (startsWith($attr_val, array('data:', 'javascript:', 'mailto:', '.'))) {
continue;
}
// Turn protocol relative URLs into HTTPS.
if (isset($attr_val[1]) && $attr_val[1] == '/' && $attr_val[0] == '/') {
$attr_val = 'https:' . $attr_val;
$element->setAttribute($attribute, $attr_val);
} else {
$attr_val_components = parse_url($attr_val);
// Nothing to do without paths.
if (isset($attr_val_components['path'])) {
// Remove current host.
if (isset($attr_val_components['host']) && $attr_val_components['host'] == $request->getUrlComponent('host')) {
unset($attr_val_components['host']);
}
// If URL without host.
if (!isset($attr_val_components['host'])) {
// If path does not start with a slash, prepend current path directory.
if ($attr_val_components['path'][0] != '/') {
$attr_val_components['path'] = dirname($request->getUrlComponent('path')) . '/' . $attr_val_components['path'];
}
$attr_val = '.' . http_build_path_query_fragment($attr_val_components);
$element->setAttribute($attribute, $attr_val);
}
}
}
/*
* Special for existing base href values. They should always end with a slash,
* but browsers are lenient if it's eg http://example.com. But since we may rewrite
* the value later, it might have a path and without the slash it will fail.
*/
if ($element->tag == 'base') {
$last_char = $attr_val[strlen($attr_val) - 1];
if ($last_char != '/') {
$attr_val .= '/';
$element->setAttribute($attribute, $attr_val);
}
}
}
}
}
public function getUrl()
{
static $url;
if (!isset($url)) {
if (isset($_GET[RedirectWhenBlockedFull::QUERY_STRING_PARAM_NAME]) && $_GET[RedirectWhenBlockedFull::QUERY_STRING_PARAM_NAME] == Conf::OUTPUT_TYPE_APK && Conf::$apk_url) {
$url = Conf::$apk_url;
$filename = basename(parse_url($url, PHP_URL_PATH));
header('Content-Disposition: attachment; filename=' . $filename);
// Run after all other code to override other content-type header.
register_shutdown_function(function () {
header('Content-Type: application/vnd.android.package-archive');
});
} else {
$url = RedirectWhenBlockedFull::getRequestUriWithoutQueryStringParam();
$this->removeThisScriptDirFromUrl($url);
if (startsWith($url, '/http://') || startsWith($url, '/https://')) {
$url = substr($url, 1);
if (!TextExternalUrlFilters::matchesUrl($url)) {
header('HTTP/1.0 403 Forbidden');
exit;
}
// If we for some reason have the default upstream host and scheme in the URL, remove them.
$url_components = parse_url($url);
if ($url_components['host'] == Conf::getDefaultUpstreamBaseUrlComponent('host') && $url_components['scheme'] == Conf::getDefaultUpstreamBaseUrlComponent('scheme')) {
$new_url = http_build_path_query_fragment($url_components);
$new_url = RedirectWhenBlockedFull::getBaseUrl() . ltrim($new_url, '/');
header('Location: ' . $new_url);
exit;
}
// Use in DomUtlFilters for relative URLs.
$base_url_suffix = rtrim(http_build_scheme_host($url), '/') . '/';
RedirectWhenBlockedFull::setBaseUrlSuffix($base_url_suffix);
} else {
if ($url == '/') {
if (Conf::$default_upstream_url) {
$url = Conf::$default_upstream_url;
}
}
$url = Conf::$default_upstream_base_url . $url;
}
}
}
// Reverse rewrites of parameters inside URL.
TextExternalUrlFilters::applyReverse($url);
Log::add($url, 'url');
return $url;
}
