soundcloud lolllllll

2 files changed, 409 insertions, 7 deletions

diff --git a/scraper/sc.php b/scraper/sc.php
new file mode 100644
index 0000000..88d4dab
--- /dev/null
+++ b/scraper/sc.php
@@ -0,0 +1,397 @@
+<?php
+
+class sc{
+	
+	public function __construct(){
+		
+		include "lib/nextpage.php";
+		$this->nextpage = new nextpage("sc");
+	}
+	
+	public function getfilters($page){
+		
+		return [
+			"type" => [
+				"display" => "Type",
+				"option" => [
+					"any" => "Any type",
+					"track" => "Tracks",
+					"people" => "People",
+					"album" => "Albums",
+					"playlist" => "Playlists",
+					"goplus" => "Go+ Tracks"
+				]
+			]
+		];
+	}
+	
+	private function get($url, $get = []){
+		
+		$curlproc = curl_init();
+		
+		if($get !== []){
+			$get = http_build_query($get);
+			$url .= "?" . $get;
+		}
+		
+		curl_setopt($curlproc, CURLOPT_URL, $url);
+		
+		curl_setopt($curlproc, CURLOPT_ENCODING, ""); // default encoding
+		curl_setopt($curlproc, CURLOPT_HTTPHEADER,
+			["User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/116.0",
+			"Accept: application/json, text/javascript, */*; q=0.01",
+			"Accept-Language: en-US,en;q=0.5",
+			"Accept-Encoding: gzip",
+			"Referer: https://soundcloud.com/",
+			"Origin: https://soundcloud.com",
+			"DNT: 1",
+			"Connection: keep-alive",
+			"Sec-Fetch-Dest: empty",
+			"Sec-Fetch-Mode: cors",
+			"Sec-Fetch-Site: same-site"]
+		);
+		
+		curl_setopt($curlproc, CURLOPT_RETURNTRANSFER, true);
+		curl_setopt($curlproc, CURLOPT_SSL_VERIFYHOST, 2);
+		curl_setopt($curlproc, CURLOPT_SSL_VERIFYPEER, true);
+		curl_setopt($curlproc, CURLOPT_CONNECTTIMEOUT, 30);
+		curl_setopt($curlproc, CURLOPT_TIMEOUT, 30);
+		
+		$data = curl_exec($curlproc);
+		
+		if(curl_errno($curlproc)){
+			
+			throw new Exception(curl_error($curlproc));
+		}
+		
+		curl_close($curlproc);
+		return $data;
+	}
+	
+	public function music($get){
+		
+		if($get["npt"]){
+			
+			$params = $this->nextpage->get($get["npt"], "music");
+			$params = json_decode($params, true);
+			
+			$url = $params["url"];
+			unset($params["url"]);
+			
+		}else{
+			
+			// normal search:
+			// https://api-v2.soundcloud.com/search?q=freddie%20dredd&variant_ids=&facet=model&user_id=351062-302234-707916-795081&client_id=iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E&limit=20&offset=0&linked_partitioning=1&app_version=1693487844&app_locale=en
+			
+			// soundcloud go+ search:
+			// https://api-v2.soundcloud.com/search/tracks?q=freddie%20dredd&variant_ids=&filter.content_tier=SUB_HIGH_TIER&facet=genre&user_id=630591-269800-703400-765403&client_id=iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E&limit=20&offset=0&linked_partitioning=1&app_version=1693487844&app_locale=en
+			
+			// tracks search:
+			// https://api-v2.soundcloud.com/search/tracks?q=freddie%20dredd&variant_ids=&facet=genre&user_id=630591-269800-703400-765403&client_id=iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E&limit=20&offset=0&linked_partitioning=1&app_version=1693487844&app_locale=en
+			
+			// users search:
+			// https://api-v2.soundcloud.com/search/users?q=freddie%20dredd&variant_ids=&facet=place&user_id=630591-269800-703400-765403&client_id=iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E&limit=20&offset=0&linked_partitioning=1&app_version=1693487844&app_locale=en
+			
+			// albums search:
+			// https://api-v2.soundcloud.com/search/albums?q=freddie%20dredd&variant_ids=&facet=genre&user_id=630591-269800-703400-765403&client_id=iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E&limit=20&offset=0&linked_partitioning=1&app_version=1693487844&app_locale=en
+			
+			// playlists search:
+			// https://api-v2.soundcloud.com/search/playlists_without_albums?q=freddie%20dredd&variant_ids=&facet=genre&user_id=630591-269800-703400-765403&client_id=iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E&limit=20&offset=0&linked_partitioning=1&app_version=1693487844&app_locale=en
+			
+			$search = $get["s"];
+			$type = $get["type"];
+			
+			switch($type){
+				
+				case "any":
+					$url = "https://api-v2.soundcloud.com/search";
+					$params = [
+						"q" => $search,
+						"variant_ids" => "",
+						"facet" => "model",
+						"user_id" => "351062-302234-707916-795081",
+						"client_id" => "iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E",
+						"limit" => 20,
+						"offset" => 0,
+						"linked_partitioning" => 1,
+						"app_version" => 1693487844,
+						"app_locale" => "en"
+					];
+					break;
+				
+				case "track":
+					$url = "https://api-v2.soundcloud.com/search/tracks";
+					$params = [
+						"q" => $search,
+						"variant_ids" => "",
+						"facet_genre" => "",
+						"user_id" => "351062-302234-707916-795081",
+						"client_id" => "iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E",
+						"limit" => 20,
+						"offset" => 0,
+						"linked_partitioning" => 1,
+						"app_version" => 1693487844,
+						"app_locale" => "en"
+					];
+					break;
+				
+				case "people":
+					$url = "https://api-v2.soundcloud.com/search/users";
+					$params = [
+						"q" => $search,
+						"variant_ids" => "",
+						"facet" => "place",
+						"user_id" => "351062-302234-707916-795081",
+						"client_id" => "iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E",
+						"limit" => 20,
+						"offset" => 0,
+						"linked_partitioning" => 1,
+						"app_version" => 1693487844,
+						"app_locale" => "en"
+					];
+					break;
+				
+				case "album":
+					$url = "https://api-v2.soundcloud.com/search/albums";
+					$params = [
+						"q" => $search,
+						"variant_ids" => "",
+						"facet" => "genre",
+						"user_id" => "351062-302234-707916-795081",
+						"client_id" => "iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E",
+						"limit" => 20,
+						"offset" => 0,
+						"linked_partitioning" => 1,
+						"app_version" => 1693487844,
+						"app_locale" => "en"
+					];
+					break;
+				
+				case "playlist":
+					$url = "https://api-v2.soundcloud.com/search/playlists_without_albums";
+					$params = [
+						"q" => $search,
+						"variant_ids" => "",
+						"facet" => "genre",
+						"user_id" => "351062-302234-707916-795081",
+						"client_id" => "iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E",
+						"limit" => 20,
+						"offset" => 0,
+						"linked_partitioning" => 1,
+						"app_version" => 1693487844,
+						"app_locale" => "en"
+					];
+					break;
+				
+				case "goplus":
+					$url = "https://api-v2.soundcloud.com/search/tracks";
+					$params = [
+						"q" => $search,
+						"variant_ids" => "",
+						"filter.content_tier" => "SUB_HIGH_TIER",
+						"facet" => "genre",
+						"user_id" => "351062-302234-707916-795081",
+						"client_id" => "iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E",
+						"limit" => 20,
+						"offset" => 0,
+						"linked_partitioning" => 1,
+						"app_version" => 1693487844,
+						"app_locale" => "en"
+					];
+					break;
+			}
+		}
+			
+		try{
+			
+			$json = $this->get($url, $params);
+			
+		}catch(Exception $error){
+			
+			throw new Exception("Failed to fetch JSON");
+		}
+		
+		/*
+		$handle = fopen("scraper/soundcloud.json", "r");
+		$json = fread($handle, filesize("scraper/soundcloud.json"));
+		fclose($handle);
+		*/
+		
+		$json = json_decode($json, true);
+		
+		if($json === null){
+			
+			throw new Exception("Failed to decode JSON");
+		}
+		
+		$out = [
+			"status" => "ok",
+			"npt" => null,
+			"song" => [],
+			"playlist" => [],
+			"author" => []
+		];
+		
+		/*
+			Get next page
+		*/
+		if(isset($json["next_href"])){
+			
+			$params["query_urn"] = $json["query_urn"];
+			$params["offset"] = $params["offset"] + 20;
+			$params["url"] = $url; // we will remove this later
+			
+			$out["npt"] =
+				$this->nextpage->store(
+					json_encode($params),
+					"music"
+				);
+		}
+		
+		/*
+			Scrape items
+		*/
+		foreach($json["collection"] as $item){
+			
+			switch($item["kind"]){
+				
+				case "user":
+					// parse author
+					$out["author"][] = [
+						"title" => $item["username"],
+						"followers" => $item["followers_count"],
+						"description" => $item["track_count"] . " songs. " . $this->limitstrlen($item["description"]),
+						"thumb" => [
+							"url" => $item["avatar_url"],
+							"ratio" => "1:1"
+						],
+						"url" => $item["permalink_url"]
+					];
+					break;
+				
+				case "playlist":
+					// parse playlist
+					$description = [];
+					$count = 0;
+					
+					foreach($item["tracks"] as $song){
+						
+						$count++;
+						
+						if(!isset($song["title"])){
+							
+							continue;
+						}
+						
+						$description[] = $song["title"];
+					}
+					
+					if(count($description) != 0){
+						
+						$description = $count . " songs. " . implode(", ", $description);
+					}
+					
+					if(
+						isset($item["artwork_url"]) &&
+						!empty($item["artwork_url"])
+					){
+						
+						$thumb = [
+							"ratio" => "1:1",
+							"url" => $item["artwork_url"]
+						];
+						
+					}elseif(
+						isset($item["tracks"][0]["artwork_url"]) &&
+						!empty($item["tracks"][0]["artwork_url"])
+					){
+						
+						$thumb = [
+							"ratio" => "1:1",
+							"url" => $item["tracks"][0]["artwork_url"]
+						];
+					}else{
+						
+						$thumb = [
+							"ratio" => null,
+							"url" => null
+						];
+					}
+					
+					$out["playlist"][] = [
+						"title" => $item["title"],
+						"description" => $description,
+						"author" => [
+							"name" => $item["user"]["username"],
+							"url" => $item["user"]["permalink_url"],
+							"avatar" => $item["user"]["avatar_url"]
+						],
+						"thumb" => $thumb,
+						"date" => strtotime($item["created_at"]),
+						"duration" => $item["duration"] / 1000,
+						"url" => $item["permalink_url"]
+					];
+					break;
+				
+				case "track":
+					if(stripos($item["monetization_model"], "TIER") === false){
+						
+						$stream = [
+							"endpoint" => "audio_sc",
+							"url" =>
+								$item["media"]["transcodings"][0]["url"] .
+								"?client_id=iMxZgT5mfGstBj8GWJbYMvpzelS8ne0E" .
+								"&track_authorization=" .
+								$item["track_authorization"]
+						];
+					}else{
+						
+						$stream = [
+							"endpoint" => null,
+							"url" => null
+						];
+					}
+					
+					// parse track
+					$out["song"][] = [
+						"title" => $item["title"],
+						"description" => $item["description"] == "" ? null : $this->limitstrlen($item["description"]),
+						"url" => $item["permalink_url"],
+						"views" => $item["playback_count"],
+						"author" => [
+							"name" => $item["user"]["username"],
+							"url" => $item["user"]["permalink_url"],
+							"avatar" => $item["user"]["avatar_url"]
+						],
+						"thumb" => [
+							"ratio" => "1:1",
+							"url" => $item["artwork_url"]
+						],
+						"date" => strtotime($item["created_at"]),
+						"duration" => (int)$item["full_duration"] / 1000,
+						"stream" => $stream
+					];
+					break;
+			}
+		}
+		
+		return $out;
+	}
+	
+	private function limitstrlen($text){
+		
+		return
+			explode(
+				"\n",
+				wordwrap(
+					str_replace(
+						"\n",
+						" ",
+						$text
+					),
+					300,
+					"\n"
+				)
+			)[0];
+	}
+}
diff --git a/scraper/yandex.php b/scraper/yandex.php
index 8cb733e..65abe73 100644
--- a/scraper/yandex.php
+++ b/scraper/yandex.php
@@ -959,6 +959,7 @@ class yandex{
 					"img"
 				);
 			
+			$c = 1;
 			if(count($thumb) === 0){
 				
 				$thumb = [
@@ -967,7 +968,6 @@ class yandex{
 				];
 			}else{
 				
-				$c = 1;
 				$thumb = [
 					"url" =>
 						str_replace(
@@ -1065,12 +1065,17 @@ class yandex{
 				"views" => $views,
 				"thumb" => $thumb,
 				"url" =>
-					$this->fuckhtml
-					->getTextContent(
-						$data["counters"]
-						["toHostingLoaded"]
-						["postfix"]
-						["href"]
+					str_replace(
+						"http://",
+						"https://",
+						$this->fuckhtml
+						->getTextContent(
+							$data["counters"]
+							["toHostingLoaded"]
+							["postfix"]
+							["href"]
+						),
+						$c
 					)
 			];
 		}