lolcat/4get
5
21
Fork 0
Code Issues 27 Pull Requests 1 Actions Packages Projects Releases Wiki Activity

bypass google soft captcha

Browse Source
This commit is contained in:
lolcat
2026-08-06 01:21:27 -04:00
parent 4ca48d0b61
commit 53e315787c
1 changed files with 80 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

97
scraper/google.php
View File

@@ -721,19 +721,19 @@ class google{
$params["tbs"] = rtrim($params["tbs"], ","); $params["tbs"] = rtrim($params["tbs"], ",");
} }
/*
$data = $this->get( $data = $this->get(
$proxy, $proxy,
"https://www.google.com/search", "https://www.google.com/search",
$params, $params,
null null
); );*/
$html = &$data["data"]; $html = &$data["data"];
$container = &$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/google-softcaptcha.html"); $html = file_get_contents("scraper/google-softcaptcha.html");
} }
$out = [ $out = [
@@ -756,7 +756,22 @@ class google{
// init // init
$this->parsestyles(); $this->parsestyles();
$this->detect_sorry();
// solve softcaptcha
$softcaptcha = $this->detect_sorry();
if($softcaptcha){
$params["sei"] = $softcaptcha["sei"];
$data = $this->get(
$proxy,
"https://www.google.com/search",
$params,
$container
);
}
$this->scrape_dimg($html); $this->scrape_dimg($html);
$this->scrape_imagearr($html); $this->scrape_imagearr($html);
@@ -1384,7 +1399,19 @@ class google{
// parse all <style> tags // parse all <style> tags
$this->parsestyles(); $this->parsestyles();
$this->detect_sorry(); $softcaptcha = $this->detect_sorry();
if($softcaptcha){
$params["sei"] = $softcaptcha["sei"];
$data = $this->get(
$proxy,
"https://www.google.com/search",
$params,
$container
);
}
// get javascript images // get javascript images
$this->scrape_imagearr($html); $this->scrape_imagearr($html);
@@ -1572,7 +1599,19 @@ class google{
// //
// Parse web video page // Parse web video page
// //
$this->detect_sorry(); $softcaptcha = $this->detect_sorry();
if($softcaptcha){
$params["sei"] = $softcaptcha["sei"];
$data = $this->get(
$proxy,
"https://www.google.com/search",
$params,
$container
);
}
// get javascript images // get javascript images
$this->scrape_dimg($html); $this->scrape_dimg($html);
@@ -1893,7 +1932,19 @@ class google{
// parse styles // parse styles
$this->parsestyles(); $this->parsestyles();
$this->detect_sorry(); $softcaptcha = $this->detect_sorry();
if($softcaptcha){
$params["sei"] = $softcaptcha["sei"];
$data = $this->get(
$proxy,
"https://www.google.com/search",
$params,
$container
);
}
// get images // get images
$this->scrape_dimg($html); $this->scrape_dimg($html);
@@ -2687,22 +2738,34 @@ class google{
throw new Exception("Google returned a JS challenge"); throw new Exception("Google returned a JS challenge");
} }
// detect soft captcha // bypass soft captcha
// "Verifying your request // "Verifying your request
// You'll be able to continue in a few seconds. Don't refresh this page." // You'll be able to continue in a few seconds. Don't refresh this page."
$scripts =
// getting this mostly on mid-trust IPs
$softcaptcha =
$this->fuckhtml $this->fuckhtml
->getElementsByAttributeValue( ->getElementsByTagName(
"role", "script"
"progressbar",
"div"
); );
if(count($softcaptcha) !== 0){ foreach($scripts as $sc){
throw new Exception("Google returned a soft captcha"); if(
preg_match(
'/var eid ?= ?\'(.*)\'[\S\s]*, ?([0-9]+)\);/U',
$sc["innerHTML"],
$matches
)
){
sleep(10); // gotta sleep to not trigger captcha. blame google, not me
return [
"sei" => $matches[1],
"timeout" => (int)$matches[2] // set to 10000 usually
];
} }
} }
return false; // no need to bypass
}
} }