diff options
| author | lolcat <will@lolcat.ca> | 2024-03-20 10:59:51 -0400 |
|---|---|---|
| committer | lolcat <will@lolcat.ca> | 2024-03-20 10:59:51 -0400 |
| commit | 635b2386d4845d7360ea5d1692090dae1dc02e68 (patch) | |
| tree | 860bdb1f8e2b82e91c64dc257dda862890217db1 | |
| parent | c4c008c1921f2efd8019d96af4d1897cc817c18c (diff) | |
fixed google not working in yurop
| -rw-r--r-- | data/config.php | 2 | ||||
| -rw-r--r-- | scraper/google.php | 34 |
2 files changed, 30 insertions, 6 deletions
diff --git a/data/config.php b/data/config.php index fd9071e..0a73864 100644 --- a/data/config.php +++ b/data/config.php @@ -92,7 +92,7 @@ class config{ const PROXY_DDG = false; // duckduckgo const PROXY_BRAVE = false; const PROXY_FB = false; // facebook - const PROXY_GOOGLE = false; + const PROXY_GOOGLE = "webshare_google"; const PROXY_MARGINALIA = false; const PROXY_MOJEEK = false; const PROXY_SC = false; // soundcloud diff --git a/scraper/google.php b/scraper/google.php index d177424..1485436 100644 --- a/scraper/google.php +++ b/scraper/google.php @@ -522,6 +522,7 @@ class google{ "Accept-Language: en-US,en;q=0.5", "Accept-Encoding: gzip", "DNT: 1", + "Cookie: SOCS=CAESNQgCEitib3FfaWRlbnRpdHlmcm9udGVuZHVpc2VydmVyXzIwMjQwMzE3LjA4X3AwGgJlbiAEGgYIgM7orwY", "Connection: keep-alive", "Upgrade-Insecure-Requests: 1", "Sec-Fetch-Dest: document", @@ -977,9 +978,9 @@ class google{ "related" => [] ]; - if($this->detect_sorry($html)){ + if($error = $this->detect_sorry($html)){ - throw new Exception("Google blocked this 4get instance. Please set up a proxy!"); + throw new Exception($error); } $this->parsejavascript($html); @@ -2813,9 +2814,9 @@ class google{ throw new Exception("Failed to get search page"); } - if($this->detect_sorry($html)){ + if($error = $this->detect_sorry($html)){ - throw new Exception("Google blocked this 4get instance. Please set up a proxy!"); + throw new Exception($error); } $out = [ @@ -3649,7 +3650,30 @@ class google{ $detect_sorry[0]["innerHTML"] == "302 Moved" ){ - return true; + // may be consent.google.com in europe or /sorry captcha page + $url = + $this->fuckhtml + ->getElementsByTagName("a"); + + if( + strpos( + parse_url( + $this->fuckhtml + ->getTextContent( + $url[0]["attributes"]["href"] + ), + PHP_URL_PATH + ), + "/sorry" + ) === 0 + ){ + + // found /sorry + return "Google blocked this 4get instance. Please setup a proxy!"; + } + + // found consent.google, should not happen anymore + return "Google served a GPDR consent form. This should not happen, please report if you encounter this message"; } return false; |