public static function run()
{
/*
* Normal request. Substitute the response with our own page.
*/
if (!isset($_GET[self::QUERY_STRING_PARAM_NAME])) {
$iframe_src = $_SERVER['REQUEST_URI'];
if ($_SERVER['QUERY_STRING']) {
$iframe_src .= '&';
} else {
$iframe_src .= '?';
}
$iframe_src .= self::QUERY_STRING_PARAM_NAME . '=' . self::OUTPUT_TYPE_IFRAME;
$request_path_depth = count(explode('/', parse_url($_SERVER['REQUEST_URI'], PHP_URL_PATH)));
$script_path_depth = count(explode('/', $_SERVER['SCRIPT_NAME']));
$rwb_path_relative_to_request_path = str_repeat('../', $request_path_depth - $script_path_depth) . 'rwb';
require 'substitute-page.php';
exit;
}
if (self::getOutputType() == self::OUTPUT_TYPE_JSONP) {
// Output header now since other header output might block it later.
header('Content-Type: application/javascript');
}
// Turn on output buffer to capture all output.
ob_start();
// Make this run after all output is completed.
register_shutdown_function(function () {
$html = ob_get_clean();
RedirectWhenBlockedFull::injectBaseTag($html);
/*
* This request comes from another base url (mirror or source host).
* We take the normal output and turn it into a jsonp response.
*/
if (RedirectWhenBlockedFull::getOutputType() == RedirectWhenBlockedFull::OUTPUT_TYPE_JSONP) {
print self::getJsonpCallbackName() . '(' . json_encode(array('html' => mb_convert_encoding_plus($html, 'UTF-8', RedirectWhenBlockedFull::getCharset($html)))) . ')';
} else {
print $html;
}
});
}
private function getBodyFilteredByContentType($body, $content_type)
{
switch ($content_type) {
case self::CONTENT_TYPE_TEXT_CSS:
$css_base_url = $this->getCssBaseUrl();
// Convert protocol relative URLs to HTTPS.
$body = str_replace('url(//', 'url(https://', $body);
$body = str_replace('url("//', 'url("https://', $body);
$body = str_replace('url(\'//', 'url(\'https://', $body);
// Disabling this part. Not sure why it was added in the first place.
/*
* $body = str_replace('url("../', 'url("' . $css_base_url . '../',
* $body);
* $body = str_replace('url(\'../',
* 'url(\'' . $css_base_url . '../', $body);
* $body = str_replace('url(../', 'url(' . $css_base_url . '../',
* $body);
*/
$body = str_replace('url("/', 'url("' . $css_base_url, $body);
$body = str_replace('url(\'/', 'url(\'' . $css_base_url, $body);
$body = str_replace('url(/', 'url(' . $css_base_url, $body);
break;
case self::CONTENT_TYPE_TEXT_HTML:
$dom = $this->getDomFromHtml($body);
if ($dom) {
foreach ($dom->find('head title') as $title_element) {
$title = mb_convert_encoding_plus($title_element->text(), 'UTF-8', $this->getCharsetFromHeadersOrDom($dom));
if ($title) {
$title .= ' | 免翻墙镜像';
// Update title in DOM.
$title_element->innertext = mb_convert_encoding_plus($title, $this->getCharsetFromHeadersOrDom($dom), 'UTF-8');
break;
}
}
}
// Default title.
if (!isset($title) || !$title) {
$title = '免翻墙镜像';
}
// Special - override meta redirects since RWB would just slow them down.
if ($dom && count($dom->find('<META[http-equiv=refresh]'))) {
foreach ($dom->find('<META[http-equiv=refresh]') as $meta) {
$content = $meta->getAttribute('content');
if ($content) {
$content_chunks = explode(';', $content, 2);
if (isset($content_chunks[1])) {
$url_chunks = explode('=', $content_chunks[1], 2);
if (isset($url_chunks[1])) {
$url = trim($url_chunks[1], '\'" ');
if ($url) {
$this->headers['Location'] = $url;
return '';
}
}
}
}
}
}
if ($this->shouldUseRedirectWhenBlocked()) {
RedirectWhenBlockedFull::setCharset($this->getCharsetFromHeadersOrDom($dom));
RedirectWhenBlockedFull::setWebsiteTitle($title);
RedirectWhenBlockedFull::run();
if (RedirectWhenBlockedFull::getOutputType() == RedirectWhenBlockedFull::OUTPUT_TYPE_JSONP) {
// RWB will turn output into JSONP. Remove content-type header to keep header sent by RWB.
if (isset($this->headers['Content-Type'])) {
unset($this->headers['Content-Type']);
}
// Also remove charset specification in inline HTML.
if ($dom) {
foreach ($dom->find('meta[http-equiv=Content-Type], meta[charset]') as $meta) {
$meta->outertext = '';
}
}
}
}
// Might be for example JSONP data served with an incorrect header.
if (!$dom || !$dom->find('body')) {
break;
}
DomUrlFilters::applyAll($dom, $this->request);
// Special for iframes. Prevent RWB from hijacking content (iframe in iframe problem).
foreach ($dom->find('iframe') as $iframe) {
$src = $iframe->getAttribute('src');
if ($src) {
$src = http_add_query_component($src, RedirectWhenBlockedFull::QUERY_STRING_PARAM_NAME, RedirectWhenBlockedFull::OUTPUT_TYPE_IFRAME);
$iframe->setAttribute('src', $src);
}
}
// Special for style attributes - apply CSS filter.
foreach ($dom->find('*[style]') as $style_element) {
$style_element->setAttribute('style', $this->getBodyFilteredByContentType($style_element->getAttribute('style'), self::CONTENT_TYPE_TEXT_CSS));
}
$dom_do_reset = false;
// Don't insert stuff if request failed.
if ($this->getResponseCode() == 200) {
if (Conf::$html_body_appendix) {
foreach ($dom->find('body') as $b) {
$b->innertext .= Conf::$html_body_appendix;
break;
}
$dom_do_reset = true;
}
if (Conf::$html_head_appendix) {
foreach ($dom->find('head') as $h) {
$h->innertext .= Conf::$html_head_appendix;
break;
}
$dom_do_reset = true;
}
}
if (count($dom->find('script[async]'))) {
if ($dom_do_reset) {
// Bug in Simple HTML Dom means that DOM has to be reloaded before further edits.
$this->resetDom($dom);
$dom_do_reset = false;
}
// Remove 'async' attributes because they break the page in IE11 when loaded in our iframe.
foreach ($dom->find('script[async]') as $x) {
$x->removeAttribute('async', '');
}
}
if (count($dom->find('style'))) {
if ($dom_do_reset) {
// Bug in Simple HTML Dom means that DOM has to be reloaded before further edits.
$this->resetDom($dom);
$dom_do_reset = false;
}
foreach ($dom->find('style') as $style) {
$style->innertext = $this->getBodyFilteredByContentType($style->innertext, self::CONTENT_TYPE_TEXT_CSS);
}
}
$body = $dom->__toString();
$this->destroyDom($dom);
break;
}
return $body;
}
