Skip to content

Commit f182fd0

Browse files
Athlon1600Athlon1600
committed
proxify plugin now detects meta refresh
1 parent 726f8a4 commit f182fd0

File tree

1 file changed

+52
-60
lines changed

1 file changed

+52
-60
lines changed

‎src/Plugin/ProxifyPlugin.php‎

Lines changed: 52 additions & 60 deletions
Original file line numberDiff line numberDiff line change
@@ -22,39 +22,27 @@ private function css_url($matches){
2222
return str_replace($matches[1], proxify_url($matches[1], $this->base_url), $matches[0]);
2323
}
2424

25-
/*
26-
27-
this.params.logoImg&&(e="background-image: url("+this.params.logoImg+")")
28-
29-
*/
25+
// this.params.logoImg&&(e="background-image: url("+this.params.logoImg+")")
3026
private function css_import($matches){
3127
return str_replace($matches[2], proxify_url($matches[2], $this->base_url), $matches[0]);
3228
}
3329

34-
private function html_href($matches){
30+
// replace src= and href=
31+
private function html_attr($matches){
3532

33+
// could be empty?
3634
$url = trim($matches[2]);
3735

38-
// do not proxify magnet: links
39-
if(strpos($url, "magnet") === 0){
36+
if(stripos($url, 'data:') === 0 || stripos($url, 'magnet:') === 0 ){
4037
return $matches[0];
4138
}
4239

43-
// do we even need to proxify this URL?
4440
return str_replace($url, proxify_url($url, $this->base_url), $matches[0]);
4541
}
4642

47-
private function html_src($matches){
48-
49-
if(stripos(trim($matches[2]), 'data:') === 0){
50-
return $matches[0];
51-
}
52-
53-
return str_replace($matches[2], proxify_url($matches[2], $this->base_url), $matches[0]);
54-
}
55-
5643
private function form_action($matches){
57-
44+
45+
// sometimes form action is empty - which means a postback to the current page
5846
// $matches[1] holds single or double quote - whichever was used by webmaster
5947

6048
// $matches[2] holds form submit URL - can be empty which in that case should be replaced with current URL
@@ -105,35 +93,59 @@ public function onBeforeRequest(ProxyEvent $event){
10593
$request->prepare();
10694
}
10795
}
96+
97+
private function meta_refresh($matches){
98+
$url = $matches[2];
99+
return str_replace($url, proxify_url($url, $this->base_url), $matches[0]);
100+
}
101+
102+
// <title>, <base>, <link>, <style>, <meta>, <script>, <noscript>
103+
private function proxify_head($str){
108104

109-
/*
110-
TODO:
111-
$input = preg_replace('#<meta[^>]*name=["\'](title|description|keywords)["\'][^>]*>#is', '', $input, 3);
112-
$input = preg_replace('#<link[^>]*rel=["\'](icon|shortcut icon)["\'][^>]*>#is', '', $input, 2);
113-
114-
# Remove and record a <base> href
115-
$input = preg_replace_callback('#<base href\s*=\s*([\\\'"])?((?(1)(?(?<=")[^"]{1,2048}|[^\\\']{1,2048})|[^\s"\\\'>]{1,2048}))(?(1)\\1|)[^>]*>#i', 'html_stripBase', $input, 1);
105+
// let's replace page titles with something custom
106+
if(Config::get('replace_title')){
107+
$str = preg_replace('/<title[^>]*>(.*?)<\/title>/is', '<title>'.Config::get('replace_title').'</title>', $str);
108+
}
116109

117-
# Proxy url= values in meta redirects
118-
$input = preg_replace_callback('#content\s*=\s*(["\\\'])?[0-9]+\s*;\s*url=([\\\'"]|&\#39;)?((?(?<=")[^"]+|(?(?<=\\\')[^\\\']+|[^\\\'" >]+)))(?(2)\\2|)(?(1)\\1|)#i', 'html_metaRefresh', $input, 1);
119110

111+
// base - update base_url contained in href - remove <base> tag entirely
112+
//$str = preg_replace_callback('/<base[^>]*href=
120113

114+
// link - replace href with proxified
115+
// link rel="shortcut icon" - replace or remove
121116

122-
# Process forms
123-
$input = preg_replace_callback('#<form([^>]*)>(.*?)</form>#is', 'html_form', $input);
117+
// meta - only interested in http-equiv - replace url refresh
118+
// <meta http-equiv="refresh" content="5; url=http://example.com/">
119+
$str = preg_replace_callback('/content=(["\'])\d+\s*;\s*url=(.*?)\1/is', array($this, 'meta_refresh'), $str);
124120

125-
*/
121+
return $str;
122+
}
123+
124+
// The <body> background attribute is not supported in HTML5. Use CSS instead.
125+
private function proxify_css($str){
126+
127+
// The HTML5 standard does not require quotes around attribute values.
128+
129+
// if {1} is not there then youtube breaks for some reason
130+
$str = preg_replace_callback('@[^a-z]{1}url\s*\((?:\'|"|)(.*?)(?:\'|"|)\)@im', array($this, 'css_url'), $str);
131+
132+
// https://developer.mozilla.org/en-US/docs/Web/CSS/@import
133+
// TODO: what about @import directives that are outside <style>?
134+
$str = preg_replace_callback('/@import (\'|")(.*?)\1/i', array($this, 'css_import'), $str);
135+
136+
return $str;
137+
}
126138

127139
public function onCompleted(ProxyEvent $event){
128140

129141
// to be used when proxifying all the relative links
130142
$this->base_url = $event['request']->getUri();
131143

132144
$response = $event['response'];
133-
$str = $response->getContent();
134-
135145
$content_type = $response->headers->get('content-type');
136146

147+
$str = $response->getContent();
148+
137149
// DO NOT do any proxification on .js files
138150
if($content_type == 'text/javascript' || $content_type == 'application/javascript' || $content_type == 'application/x-javascript'){
139151
return;
@@ -147,44 +159,24 @@ public function onCompleted(ProxyEvent $event){
147159
foreach($js_remove as $pattern){
148160
if(strpos($domain, $pattern) !== false){
149161
$str = Html::remove_scripts($str);
150-
break;
151162
}
152163
}
153164
}
154165

166+
// add html.no-js
167+
155168
// let's remove all frames?? does not protect against the frames created dynamically via javascript
156169
$str = preg_replace('@<iframe[^>]*>[^<]*<\\/iframe>@is', '', $str);
157170

158-
// let's replace page titles with something custom
159-
if(Config::get('replace_title')){
160-
$str = preg_replace('/<title[^>]*>(.*?)<\/title>/ims', '<title>'.Config::get('replace_title').'</title>', $str);
161-
}
171+
$str = $this->proxify_head($str);
172+
$str = $this->proxify_css($str);
162173

163-
/* css
164-
if {1} is not there then youtube breaks for some reason
165-
*/
166-
$str = preg_replace_callback('@[^a-z]{1}url\s*\((?:\'|"|)(.*?)(?:\'|"|)\)@im', array($this, 'css_url'), $str);
167-
168-
// https://developer.mozilla.org/en-US/docs/Web/CSS/@import
169-
// TODO: what about @import directives that are outside <style>?
170-
$str = preg_replace_callback('/@import (\'|")(.*?)\1/i', array($this, 'css_import'), $str);
171-
172-
// html .*? just in case href is empty...
173-
$str = preg_replace_callback('@href\s*=\s*(["\'])(.*?)\1@im', array($this, 'html_href'), $str);
174+
// src= and href=
175+
$str = preg_replace_callback('@(?:src|href)\s*=\s*(["|\'])(.*?)\1@is', array($this, 'html_attr'), $str);
174176

175-
176-
/*
177-
178-
src= can be empty - then what?
179-
180-
*/
181-
$str = preg_replace_callback('@src\s*=\s*(["|\'])(.*?)\1@i', array($this, 'html_src'), $str);
182-
183-
// sometimes form action is empty - which means a postback to the current page
177+
// form
184178
$str = preg_replace_callback('@<form[^>]*action=(["\'])(.*?)\1[^>]*>@i', array($this, 'form_action'), $str);
185179

186-
//$str = str_replace('document.forms[0]', 'document.forms[1]', $str);
187-
188180
$response->setContent($str);
189181
}
190182

0 commit comments

Comments
 (0)