|
@@ -875,7 +875,7 @@ class google{
|
|
|
$container = &$data["container"];
|
|
$container = &$data["container"];
|
|
|
|
|
|
|
|
//file_put_contents("scraper/google-jp.html", $html);
|
|
//file_put_contents("scraper/google-jp.html", $html);
|
|
|
- //$html = file_get_contents("scraper/sublink.html");
|
|
|
|
|
|
|
+ //$html = file_get_contents("scraper/google.html");
|
|
|
//$container = "";
|
|
//$container = "";
|
|
|
}
|
|
}
|
|
|
|
|
|
|
@@ -1048,6 +1048,8 @@ class google{
|
|
|
$this->fuckhtml
|
|
$this->fuckhtml
|
|
|
->getElementsByTagName("a");
|
|
->getElementsByTagName("a");
|
|
|
|
|
|
|
|
|
|
+ $illegal = false;
|
|
|
|
|
+
|
|
|
foreach($as as $a){
|
|
foreach($as as $a){
|
|
|
|
|
|
|
|
if(!isset($a["attributes"]["href"])){ continue; }
|
|
if(!isset($a["attributes"]["href"])){ continue; }
|
|
@@ -1068,9 +1070,23 @@ class google{
|
|
|
str_contains(
|
|
str_contains(
|
|
|
$link_text,
|
|
$link_text,
|
|
|
"complaint"
|
|
"complaint"
|
|
|
|
|
+ ) ||
|
|
|
|
|
+ str_contains(
|
|
|
|
|
+ $link_text,
|
|
|
|
|
+ "read more about the request"
|
|
|
)
|
|
)
|
|
|
){
|
|
){
|
|
|
|
|
|
|
|
|
|
+ if(
|
|
|
|
|
+ str_contains(
|
|
|
|
|
+ $link_text,
|
|
|
|
|
+ "read more about the request"
|
|
|
|
|
+ )
|
|
|
|
|
+ ){
|
|
|
|
|
+
|
|
|
|
|
+ $illegal = true;
|
|
|
|
|
+ }
|
|
|
|
|
+
|
|
|
// sometimes the link is an encrypted redirect, handle redirection securely
|
|
// sometimes the link is an encrypted redirect, handle redirection securely
|
|
|
$notices[] = $this->redirect_add_url($link);
|
|
$notices[] = $this->redirect_add_url($link);
|
|
|
}
|
|
}
|
|
@@ -1083,13 +1099,24 @@ class google{
|
|
|
|
|
|
|
|
$i = 0;
|
|
$i = 0;
|
|
|
$c = count($notices);
|
|
$c = count($notices);
|
|
|
-
|
|
|
|
|
- $notices_dom = [
|
|
|
|
|
- [
|
|
|
|
|
- "type" => "text",
|
|
|
|
|
- "value" => "Google removed results from this page, but you can get them anyways by consulting these takedown notices: "
|
|
|
|
|
- ]
|
|
|
|
|
- ];
|
|
|
|
|
|
|
+
|
|
|
|
|
+ if($illegal === false){
|
|
|
|
|
+
|
|
|
|
|
+ $notices_dom = [
|
|
|
|
|
+ [
|
|
|
|
|
+ "type" => "text",
|
|
|
|
|
+ "value" => "Google removed results from this page, but you can get them anyways by consulting these takedown notices: "
|
|
|
|
|
+ ]
|
|
|
|
|
+ ];
|
|
|
|
|
+ }else{
|
|
|
|
|
+
|
|
|
|
|
+ $notices_dom = [
|
|
|
|
|
+ [
|
|
|
|
|
+ "type" => "text",
|
|
|
|
|
+ "value" => "Google removed results from this page because they were illegal. You can view the takedown requests here: "
|
|
|
|
|
+ ]
|
|
|
|
|
+ ];
|
|
|
|
|
+ }
|
|
|
|
|
|
|
|
foreach($notices as $n){
|
|
foreach($notices as $n){
|
|
|
|
|
|