diff options
| -rw-r--r-- | lib/backend.php | 1 | ||||
| -rw-r--r-- | lib/fuckhtml.php | 3 | ||||
| -rw-r--r-- | scraper/brave.php | 23 |
3 files changed, 22 insertions, 5 deletions
diff --git a/lib/backend.php b/lib/backend.php index 09a4267..00161a6 100644 --- a/lib/backend.php +++ b/lib/backend.php @@ -74,6 +74,7 @@ class backend{ break; case "socks5_hostname": + case "socks5a": curl_setopt($curlproc, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5_HOSTNAME); curl_setopt($curlproc, CURLOPT_PROXY, $address . ":" . $port); break; diff --git a/lib/fuckhtml.php b/lib/fuckhtml.php index 8b5eba5..d89a7bc 100644 --- a/lib/fuckhtml.php +++ b/lib/fuckhtml.php @@ -424,8 +424,6 @@ class fuckhtml{ $escape = false; } - $lastchar = $json[$i]; - if( $bracket === false && $is_close_bracket === false @@ -436,7 +434,6 @@ class fuckhtml{ case "[": case "{": - // dont execute whats in "default" $json_out .= $json[$i]; break; diff --git a/scraper/brave.php b/scraper/brave.php index 8be55ac..fd4c46e 100644 --- a/scraper/brave.php +++ b/scraper/brave.php @@ -295,6 +295,7 @@ class brave{ $html = fread($handle, filesize("scraper/brave.html")); fclose($handle); */ + try{ $html = $this->get( @@ -406,9 +407,9 @@ class brave{ if(!isset($grep[1])){ - throw new Exception("Could not get data JS"); + throw new Exception("Could not grep JavaScript object"); } - + $data = $this->fuckhtml ->parseJsObject( @@ -416,6 +417,24 @@ class brave{ ); unset($grep); + if($data === null){ + + throw new Exception("Failed to decode JavaScript object"); + } + + if( + isset($data[2]["data"]["title"]) && + stripos($data[2]["data"]["title"], "PoW Captcha") !== false + ){ + + throw new Exception("Brave returned a PoW captcha"); + } + + if(!isset($data[1]["data"]["body"]["response"])){ + + throw new Exception("Brave did not return a result object"); + } + $data = $data[1]["data"]["body"]["response"]; /* |