merge brave scraper fix from upstream

f9f3c919d6

Signed-off-by: ngn <ngn@ngn.tf>

src/scraper/brave.php

@@ -209,9 +209,66 @@ class brave{
 		curl_close($curlproc);
 		return $data;
 	}
-	
+
+	private function get_js(){
+
+		$script_disc =
+			$this->fuckhtml
+			->getElementsByTagName(
+				"script"
+			);
+
+		$data = null;
+		foreach($script_disc as &$discs){
+
+			if(
+				preg_match(
+					'/kit\.start\(/',
+					$discs["innerHTML"]
+				)
+			){
+
+				$data =
+					explode(
+						"data:",
+						$discs["innerHTML"],
+						2
+					);
+
+				if(count($data) !== 2){
+
+					throw new Exception("Failed to split up data field");
+				}
+
+				$data = $data[1];
+				break;
+			}
+		}
+
+		if($data === null){
+
+			throw new Exception("Could not grep JavaScript object");
+		}
+
+		$data =
+			$this->fuckhtml
+			->parseJsObject(
+				$this->fuckhtml
+				->extract_json(
+					$data
+				)
+			);
+
+		if($data === null){
+
+			throw new Exception("Failed to decode JavaScript object");
+		}
+
+		return $data;
+	}
+
 	public function web($get){
-		
+
 		if($get["npt"]){
 			
 			// get next page data
@@ -381,57 +438,11 @@ class brave{
 				}
 			}
 		}
-		
+
+		// do some magic
 		$this->fuckhtml->load($html);
-		
+		$data = $this->get_js();
-		$script_disc =
+
-			$this->fuckhtml
-			->getElementsByTagName(
-				"script"
-			);
-		
-		$grep = [];
-		foreach($script_disc as $discs){
-			
-			preg_match(
-				'/const data ?= ?(\[{.*}]);/',
-				$discs["innerHTML"],
-				$grep
-			);
-			
-			if(isset($grep[1])){
-				
-				break;
-			}
-		}
-		
-		if(!isset($grep[1])){
-			
-			throw new Exception("Could not grep JavaScript object");
-		}
-		
-		$data =
-			rtrim(
-				preg_replace(
-					'/\(Array\(0\)\)\).*$/',
-					"",
-					$grep[1]
-				),
-				" ]"
-			) . "]";
-		
-		$data =
-			$this->fuckhtml
-			->parseJsObject(
-				$data
-			);
-		unset($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
@@ -1178,25 +1189,10 @@ class brave{
 				"news",
 				$proxy
 			);
-		
+
-		preg_match(
+		$this->fuckhtml->load($html);
-			'/const data ?= ?(\[{.*}]);/',
+		$json = $this->get_js();
-			$html,
+
-			$json
-		);
-		
-		if(!isset($json[1])){
-			
-			throw new Exception("Failed to grep javascript object");
-		}
-		
-		$json = $this->fuckhtml->parseJsObject($json[1], true);
-		
-		if($json === null){
-			
-			throw new Exception("Failed to parse javascript object");
-		}
-		
 		foreach(
 			$json[1]["data"]["body"]["response"]["news"]["results"]
 			as $news
@@ -1276,24 +1272,10 @@ class brave{
 		$handle = fopen("scraper/brave-image.html", "r");
 		$html = fread($handle, filesize("scraper/brave-image.html"));
 		fclose($handle);*/
-		
+
-		preg_match(
+		$this->fuckhtml->load($html);
-			'/const data = (\[{.*}\]);/',
+		$json = $this->get_js();
-			$html,
+
-			$json
-		);
-		
-		if(!isset($json[1])){
-			
-			throw new Exception("Failed to get data object");
-		}
-		
-		$json =
-			$this->fuckhtml
-			->parseJsObject(
-				$json[1]
-			);
-		
 		foreach(
 			$json[1]
 			["data"]
@@ -1421,24 +1403,10 @@ class brave{
 		$handle = fopen("scraper/brave-video.html", "r");
 		$html = fread($handle, filesize("scraper/brave-video.html"));
 		fclose($handle);*/
-		
+
-		preg_match(
+		$this->fuckhtml->load($html);
-			'/const data = (\[{.*}\]);/',
+		$json = $this->get_js();
-			$html,
+
-			$json
-		);
-		
-		if(!isset($json[1])){
-			
-			throw new Exception("Failed to get data object");
-		}
-		
-		$json =
-			$this->fuckhtml
-			->parseJsObject(
-				$json[1]
-			);
-		
 		foreach(
 			$json
 			[1]