ddg hotfix

Reviewed-on: lolcat/4get#41

api.txt

@@ -1,10 +1,17 @@
-                        __ __             __
-                       / // / ____ ____  / /_
-                      / // /_/ __ `/ _ \/ __/
-                     /__  __/ /_/ /  __/ /_ 
-                       /_/  \__, /\___/\__/
-                           /____/         
-
+                   44
+                 4444444      44   
+                 44444444   44444       444
+                 44444444  444444    444444444
+                  44444   44444444  444444444
+                         444444444   4444444
+                        4444444444    444444
+                      4444444444444
+                    444444444444444444
+                       444444444444444
+                          44444444
+                          4444
+                           44
+                 
            + Welcome to the 4get API documentation +
 
 + Terms of use

data/config.php

@@ -119,7 +119,7 @@ class config{
 	
 	// Default user agent to use for scraper requests. Sometimes ignored to get specific webpages
 	// Changing this might break things.
-	const USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:129.0) Gecko/20100101 Firefox/129.0";
+	const USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:133.0) Gecko/20100101 Firefox/133.0";
 	
 	// Proxy pool assignments for each scraper
 	// false = Use server's raw IP
@@ -129,6 +129,7 @@ class config{
 	const PROXY_BRAVE = false;
 	const PROXY_FB = false; // facebook
 	const PROXY_GOOGLE = false;
+	const PROXY_GOOGLE_CSE = false;
 	const PROXY_STARTPAGE = false;
 	const PROXY_QWANT = false;
 	const PROXY_GHOSTERY = false;
@@ -157,6 +158,9 @@ class config{
 	// Scraper-specific parameters
 	//
 	
+	// GOOGLE CSE
+	const GOOGLE_CX_ENDPOINT = "d4e68b99b876541f0";
+	
 	// MARGINALIA
 	// Use "null" to default out to HTML scraping OR specify a string to
 	// use the API (Eg: "public"). API has less filters.

docs/nginx.md

@@ -1,103 +1,194 @@
-# Install on NGINX
+<h1 align=center>Installation of 4get in NGINX</h1>
 
->I do NOT recommend following this guide, only follow this if you *really* need to use nginx. I recommend you use the apache2 steps instead.
+<div align=right>
 
-Login as root.
+> NOTE: As the previous version stated, it is better to follow the <a href="https://git.lolcat.ca/lolcat/4get/src/branch/master/docs/apache2.md">Apache2 guide</a> instead of the Nginx one.
 
-Create a file in `/etc/nginx/sites-avaliable/` called `4get.conf` or any name you want and put this into the file:
+> NOTE: This is going to guess that you're using either a <abbr title="(Arch Linux, Artix Linux, Endeavouros, etc...) ">Arch-based system</abbr> or a <abbr title="(Debian, Ubuntu, Devuan, etc...)">Debian-based system</abbr>, although you can still follow it with minor issues.
 
-```
-server {
-    # DO YOU REALLY NEED TO LOG SEARCHES?
-    access_log /dev/null;
-    error_log /dev/null;
-    # Change this if you have 4get in other folder.
-    root /var/www/4get;
-    # Change yourdomain by your domain lol
-    server_name www.yourdomain.com yourdomain.com;
+</div>
 
-    location @php {
-        try_files $uri.php $uri/index.php =404;
-                # Change the unix socket address if it's different for you.
-        fastcgi_pass unix:/var/run/php-fpm/php-fpm.sock;
-        fastcgi_index index.php;
-                # Change this to `fastcgi_params` if you use a debian based distro.
-        include fastcgi.conf;
-        fastcgi_intercept_errors on;
-    }
+1. Login as root.
+2. Upgrade your system:
+   * On Arch-based, run `pacman -Syu`.
+   * On Debian-based, run `apt update`, then `apt upgrade`.
+3. Install the following dependencies:
+   * `git`: So you can clone <a href="https://git.lolcat.ca/lolcat/4get">this</a> repository.
+   * `nginx`: So you can run Nginx.
+   * `php-fpm`: This is what allows Nginx to run *(and show)* PHP files.
+   * `php-imagick`, `imagemagick`: Image manipulation.
+   * `php-apcu`: Caching module.
+   * `php-curl`, `curl`:  Transferring data with URLs.
+   * `php-mbstring`: String utils.
+   * `certbot`, `certbot-nginx`: ACME client. Used to create SSL certificates.
+     * In Arch-based distributions:
+       * `pacman -S nginx certbot php-imagick certbot-nginx imagemagick curl php-apcu git`
+     * In Debian-based distributions:
+       * `apt install php-mbstring nginx certbot-nginx certbot php-imagick imagemagick php-curl curl php-apcu git`
 
-    location / {
-        try_files $uri @php;
-    }
+<div align=right>
 
-    location ~* ^(.*)\.php$ {
-        return 301 $1;
-    }
+> IMPORTANT: `php-curl`, `php-mbstring` might be a Debian-only package, but this needs further fact checking.
 
+> IMPORTANT: If having issues with `php-apcu` or `libsodium`, go to [^1].
+
+</div>
+
+4. `cd` to `/etc/nginx` and make the `conf.d/` directory if it doesn't exist:
+   * Again, this guesses you're logged in as root.
+   ```sh
+   cd /etc/nginx
+   ls -l conf.d/ # If ls shows conf.d, then it means it exists.
+   # If it does not, run:
+   mkdir conf.d
+   ```
+5. Make a file inside `conf.d/` called `4get.conf` and place the following content:
+   * First run `touch conf.d/4get.conf` then `nano conf.d/4get.conf` to open the nano editor: *(Install it if it is not, or use another editor.)*
+    ```sh
+    server {
+        access_log /dev/null; # Search log file. Do you really need to?
+        error_log /dev/null;  # Error log file.
+
+        # Change this if you have 4get in another folder.
+        root /var/www/4get;
+        # Change 'yourdomain' to your domain.
+        server_name www.yourdomain.com yourdomain.com;
+        # Port to listen to.
         listen 80;
-}
-```
 
-That is a very basic config so you will need to adapt it to your needs in case you have a more complicated nginx configuration. Anyways, you can see a real world example [here](https://git.zzls.xyz/Fijxu/etc-configs/src/branch/selfhost/nginx/sites-available/4get.zzls.xyz.conf)
+        location @php {
+            try_files $uri.php $uri/index.php =404;
+            # Change the unix socket address if it's different for you.
+            fastcgi_pass unix:/var/run/php-fpm/php-fpm.sock;
+            fastcgi_index index.php;
+            # Change this to `fastcgi_params` if you use a debian based distribution.
+            include fastcgi.conf;
+            fastcgi_intercept_errors on;
+        }
 
-After you save the file you will need to do a symlink of the `4get.conf` file to `/etc/nignx/sites-enabled/`, you can do it with this command: 
+        location / {
+            try_files $uri @php;
+        }
 
-```sh
-ln -s /etc/nginx/sites-available/4get.conf /etc/nginx/sites-available/4get.conf
-```
+        location ~* ^(.*)\.php$ {
+            return 301 $1;
+        }
 
-Now test the nginx config with `nginx -t`, if it says that everything is good, restart nginx using `systemctl restart nginx`
-
-# Encryption setup
-
-Generate a certificate for the domain using:
-
-```sh
-certbot --nginx --key-type ecdsa -d www.yourdomain.com -d yourdomain.com
-```
-(Remember to install the nginx certbot plugin!!!)
-
-After doing that certbot should deploy the certificate automatically into your 4get nginx config file. It should be ready to use at that point.
-
-# Tor setup on NGINX
-
-Important Note: Tor onion addresses are significantly longer than traditional domain names. Before proceeding with Nginx configuration, ensure you increase the `server_names_hash_bucket_size` value in your `nginx.conf` file. This setting in your Nginx configuration controls the internal data structure used to manage multiple server names (hostnames) associated with your web server. Each hostname requires a certain amount of memory within this structure. If the size is insufficient, Nginx will encounter errors.
-
-1. Open your `nginx.conf` file (that is under `/etc/nginx/nginx.conf`).
-2. Find the line containing `# server_names_hash_bucket_size 64;`.
-3. Uncomment the line and adjust the value. Start with 64, but if you encounter issues, incrementally increase it (e.g., 128, 256) until it accommodates your configuration.
-
-Open your current 4get NGINX config (that is under `/etc/nginx/sites-available/`) and append this to the end of the file:
-
-```
-server {
-	access_log /dev/null;
-	error_log /dev/null;
-
-    listen 80;
-    server_name <youronionaddress>;
-    root /var/www/4get;
-
-    location @php {
-        try_files $uri.php $uri/index.php =404;
-        # Change the unix socket address if it's different for you.
-        fastcgi_pass unix:/var/run/php-fpm/php-fpm.sock;
-        fastcgi_index index.php;
-        # Change this to `fastcgi_params` if you use a debian based distro.
-        include fastcgi.conf;
-        fastcgi_intercept_errors on;
     }
+    ```
+    * The above is a very basic configuration and thus will need tweaking to your personal needs. It should still work as-is, though. A 'real world' example is present in [^2].
+    * After saving the file, check that the `nginx.conf` file inside the main directory includes files inside `conf.d/`:
+      * It should be inside the the http block: *(The following is an example! Don't just Copy and Paste it!)*
+      ```sh
+      http {
+        include       mime.types;
+        include       conf.d/*.conf; 
+        types_hash_max_size 4096;
+        # ...
+      }
+      ```
+    * Now, test your configuration with `nginx -t`, if it says that everything is good, restart *(or start)* the Nginx daemon:
+      * This depends on the init manager, most distributions use `systemd`, but it's better practice to include most.
+      ```sh
+      # systemd
+      systemctl stop nginx
+      systemctl start nginxt
+      # or
+      systemctl restart nginx
+
+      # openrc
+      rc-service nginx stop
+      rc-service nginx start
+      # or
+      rc-service nginx restart
+
+      # runit
+      sv down nginx
+      sv up nginx
+      # or
+      sv restart nginx
+
+      # s6
+      s6-rc -d change nginx
+      s6-rc -u change nginx
+      # or
+      s6-svc -r /run/service/nginx
+
+      # dinit
+      dinitctl stop nginx
+      dinitctl start nginx
+      # or
+      dinitctl restart nginx
+      ```
+6. Clone the repository to `/var/www`:
+   * `git clone --depth 1 https://git.lolcat.ca/lolcat/4get 4get` - It clones the repository with the depth of one commit *(so it takes less time to download)* and saves the cloned repository as '4get'.
+7. That should be it! There are some extra steps you can take, but it really just depends on you.
+
+<h2 align=center>Encryption setup</h2>
+
+1. Generate a certificate for the domain you're using with:
+   * Note that `certbot-nginx` is needed.
+    ```sh
+    certbot --nginx --key-type ecdsa -d www.yourdomain.com -d yourdomain.com
+    ```
+2. After that, certbot will deploy the certificate automatically to your 4get conf file; It should be ready to use from there.
+
+<h2 align=center>Tor Setup</h2>
+
+<div align=right>
+
+> IMPORTANT: Tor onion addresses are very long compared to traditional domains, so, Before doing anything, edit `nginx.conf` and increase <abbr title="This setting in your Nginx configuration controls the internal data structure used to manage multiple server names (hostnames) associated with your web server. Each hostname requires a certain amount of memory within this structure. If the size is insufficient, Nginx will encounter errors."><code>server_names_hash_bucket_size</code></abbr> to your needs.
+
+</div>
+
+1. `cd` to `/etc/nginx` *(if you haven't)* and open your `nginx.conf` file.
+2. Find the line containing `# server_names_hash_bucket_size 64;` inside said file.
+3. Uncomment the line and adjust the value; start with 64, but if you encounter issues, incrementally increase it *(e.g., 128, 256)* until it accommodates your configuration.
+4. Open *(or duplicate the configuration)* and edit it:
+   * Example configuration, again:
+    ```sh
+    server {
+        access_log /dev/null; # Search log file. Do you really need to?
+        error_log /dev/null;  # Error log file.
+
+        # Change this if you have 4get in another folder.
+        root /var/www/4get;
+        # Change 'onionadress.onion' to your onion link.
+        server_name onionadress.onion;
+        # Port to listen to.
+        listen 80;
+
+        location @php {
+            try_files $uri.php $uri/index.php =404;
+            # Change the unix socket address if it's different for you.
+            fastcgi_pass unix:/var/run/php-fpm/php-fpm.sock;
+            fastcgi_index index.php;
+            # Change this to `fastcgi_params` if you use a debian based distribution.
+            include fastcgi.conf;
+            fastcgi_intercept_errors on;
+        }
+
+        location / {
+            try_files $uri @php;
+        }
+
+        location ~* ^(.*)\.php$ {
+            return 301 $1;
+        }
 
-    location / {
-        try_files $uri @php;
     }
+    ```
+    A real world example is present in [^2].
+5. Once done, check the configuration with `nginx -t`. If everything's fine and dandy, refer to <a href="https://git.lolcat.ca/lolcat/4get/src/branch/master/docs/tor.md">the Tor guide</a> to setup your onion site.
 
-    location ~* ^(.*)\.php$ {
-        return 301 $1;
-    }
-}
-```
+<h2 align=center>Other important things</h2>
 
-Obviously replace `<youronionaddress>` by the onion address of `/var/lib/tor/4get/hostname` and then check if the nginx config is valid with `nginx -t` if yes, then restart the nginx service and try opening the onion address into the Tor Browser. You can see a real world example [here](https://git.zzls.xyz/Fijxu/etc-configs/src/branch/selfhost/nginx/sites-available/4get.zzls.xyz.conf)
+1. <a href="https://git.lolcat.ca/lolcat/4get/src/branch/master/docs/configure.md">Configuration guide</a>: Things to do after setup.
+2. <a href="https://git.lolcat.ca/lolcat/4get/src/branch/master/docs/apache2.md">Apache2 guide</a>: Fallback to this if you couldn't get something to work, or you don't know something.
 
-Once you did the above, refer to <a href="https://git.lolcat.ca/lolcat/4get/src/branch/master/docs/tor.md">this tor guide</a> to setup your onionsite.
+<h2 align=center>Known issues</h2>
+
+1. https://git.lolcat.ca/lolcat/4get/issues
+
+[^1]: lolcat/4get#40, If having issues with `libsodium`, or `php-apcu`.
+[^2]: <a href="https://git.nadeko.net/Fijxu/etc-configs/src/branch/selfhost/nginx/conf.d/4get.conf">git.nadeko.net</a> nadeko.net's 4get instance configuration.

lib/backend.php

@@ -75,6 +75,7 @@ class backend{
 				break;
 			
 			case "socks5_hostname":
+			case "socks5h":
 			case "socks5a":
 				curl_setopt($curlproc, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5_HOSTNAME);
 				curl_setopt($curlproc, CURLOPT_PROXY, $address . ":" . $port);

lib/frontend.php

@@ -838,10 +838,10 @@ class frontend{
 		}
 		
 		$payload .=
-				'<a href="https://webcache.googleusercontent.com/search?q=cache:' . $urlencode . '" class="list" target="_BLANK"><img src="/favicon?s=https://google.com" alt="go">Google cache</a>' .
 				'<a href="https://web.archive.org/web/' . $urlencode . '" class="list" target="_BLANK"><img src="/favicon?s=https://archive.org" alt="ar">Archive.org</a>' .
 				'<a href="https://archive.ph/newest/' . htmlspecialchars($link) . '" class="list" target="_BLANK"><img src="/favicon?s=https://archive.is" alt="ar">Archive.is</a>' .
 				'<a href="https://ghostarchive.org/search?term=' . $urlencode . '" class="list" target="_BLANK"><img src="/favicon?s=https://ghostarchive.org" alt="gh">Ghostarchive</a>' .
+				'<a href="https://arquivo.pt/wayback/' . htmlspecialchars($link) . '" class="list" target="_BLANK"><img src="/favicon?s=https://arquivo.pt" alt="ar">Arquivo.pt</a>' .
 				'<a href="https://www.bing.com/search?q=url%3A' . $urlencode . '" class="list" target="_BLANK"><img src="/favicon?s=https://bing.com" alt="bi">Bing cache</a>' .
 				'<a href="https://megalodon.jp/?url=' . $urlencode . '" class="list" target="_BLANK"><img src="/favicon?s=https://megalodon.jp" alt="me">Megalodon</a>' .
 			'</div>';
@@ -939,6 +939,7 @@ class frontend{
 						"brave" => "Brave",
 						"yandex" => "Yandex",
 						"google" => "Google",
+						"google_cse" => "Google CSE",
 						"startpage" => "Startpage",
 						"qwant" => "Qwant",
 						"ghostery" => "Ghostery",
@@ -963,6 +964,7 @@ class frontend{
 						"yandex" => "Yandex",
 						"brave" => "Brave",
 						"google" => "Google",
+						"google_cse" => "Google CSE",
 						"startpage" => "Startpage",
 						"qwant" => "Qwant",
 						"yep" => "Yep",

lib/fuckhtml.php

@@ -381,6 +381,8 @@ class fuckhtml{
 		$json_out = null;
 		$last_char = null;
 		
+		$keyword_check = null;
+		
 		for($i=0; $i<strlen($json); $i++){
 			
 			switch($json[$i]){
@@ -396,6 +398,7 @@ class fuckhtml{
 						
 						$bracket = false;
 						$is_close_bracket = true;
+						
 					}else{
 						
 						if($bracket === false){
@@ -429,6 +432,31 @@ class fuckhtml{
 				$is_close_bracket === false
 			){
 				
+				// do keyword check
+				$keyword_check .= $json[$i];
+				
+				if(in_array($json[$i], [":", "{"])){
+					
+					$keyword_check = substr($keyword_check, 0, -1);
+					
+					if(
+						preg_match(
+							'/function|array|return/i',
+							$keyword_check
+						)
+					){
+						
+						$json_out =
+							preg_replace(
+								'/[{"]*' . preg_quote($keyword_check, "/") . '$/',
+								"",
+								$json_out
+							);
+					}
+					
+					$keyword_check = null;
+				}
+				
 				// here we know we're not iterating over a quoted string
 				switch($json[$i]){
 					
@@ -498,4 +526,85 @@ class fuckhtml{
 				$string
 			);
 	}
+	
+	public function extract_json($json){
+		
+		$len = strlen($json);
+		$array_level = 0;
+		$object_level = 0;
+		$in_quote = null;
+		$start = null;
+		
+		for($i=0; $i<$len; $i++){
+			
+			switch($json[$i]){
+				
+				case "[":
+					if($in_quote === null){
+						
+						$array_level++;
+						if($start === null){
+							
+							$start = $i;
+						}
+					}
+					break;
+				
+				case "]":
+					if($in_quote === null){
+						
+						$array_level--;
+					}
+					break;
+				
+				case "{":
+					if($in_quote === null){
+						
+						$object_level++;
+						if($start === null){
+							
+							$start = $i;
+						}
+					}
+					break;
+				
+				case "}":
+					if($in_quote === null){
+						
+						$object_level--;
+					}
+					break;
+				
+				case "\"":
+				case "'":
+					if(
+						$i !== 0 &&
+						$json[$i - 1] !== "\\"
+					){
+						// found a non-escaped quote
+						
+						if($in_quote === null){
+							
+							// open quote
+							$in_quote = $json[$i];
+						}elseif($in_quote === $json[$i]){
+							
+							// close quote
+							$in_quote = null;
+						}
+					}
+					break;
+			}
+			
+			if(
+				$start !== null &&
+				$array_level === 0 &&
+				$object_level === 0
+			){
+				
+				return substr($json, $start, $i - $start + 1);
+				break;
+			}
+		}
+	}
 }

scraper/brave.php

@@ -293,8 +293,8 @@ class brave{
 		/*
 		$handle = fopen("scraper/brave.html", "r");
 		$html = fread($handle, filesize("scraper/brave.html"));
-		fclose($handle);
-		*/
+		fclose($handle);*/
+		
 		
 		try{
 			$html =
@@ -410,10 +410,20 @@ class brave{
 			throw new Exception("Could not grep JavaScript object");
 		}
 		
+		$data =
+			rtrim(
+				preg_replace(
+					'/\(Array\(0\)\)\).*$/',
+					"",
+					$grep[1]
+				),
+				" ]"
+			) . "]";
+		
 		$data =
 			$this->fuckhtml
 			->parseJsObject(
-				$grep[1]
+				$data
 			);
 		unset($grep);
 		
@@ -663,7 +673,10 @@ class brave{
 					$table["Address"] = $result["location"]["postal_address"]["displayAddress"];
 				}
 				
-				if(isset($result["location"]["rating"])){
+				if(
+					isset($result["location"]["rating"]) &&
+					$result["location"]["rating"] != "void 0"
+				){
 					
 					$table["Rating"] =
 						$result["location"]["rating"]["ratingValue"] . "/" .
@@ -671,13 +684,19 @@ class brave{
 						number_format($result["location"]["rating"]["reviewCount"]) . " votes)";
 				}
 				
-				if(isset($result["location"]["contact"]["telephone"])){
+				if(
+					isset($result["location"]["contact"]["telephone"]) &&
+					$result["location"]["contact"]["telephone"] != "void 0"
+				){
 					
 					$table["Phone number"] =
 						$result["location"]["contact"]["telephone"];
 				}
 				
-				if(isset($result["location"]["price_range"])){
+				if(
+					isset($result["location"]["price_range"]) &&
+					$result["location"]["price_range"] != "void 0"
+				){
 					
 					$table["Price"] =
 						$result["location"]["price_range"];

scraper/ftm.php

@@ -136,7 +136,7 @@ class ftm{
 				"source" => [
 					[
 						"url" =>
-							"https://findthatmeme.us-southeast-1.linodeobjects.com/" .
+							"https://s3.thehackerblog.com/findthatmeme/" .
 							$thumb,
 						"width" => null,
 						"height" => null

scraper/marginalia.php

@@ -220,6 +220,7 @@ class marginalia{
 			"related" => []
 		];
 		
+		// API scraper
 		if(config::MARGINALIA_API_KEY !== null){
 			
 			try{
@@ -263,34 +264,57 @@ class marginalia{
 			return $out;
 		}
 		
-		// no more cloudflare!! Parse html by default
-		$params = [
-			"query" => $search
-		];
+		// HTML parser
+		$proxy = $this->backend->get_ip();
 		
-		foreach(["adtech", "recent", "intitle"] as $v){
+		if($get["npt"]){
 			
-			if($get[$v] == "yes"){
+			[$params, $proxy] =
+				$this->backend->get(
+					$get["npt"],
+					"web"
+				);
+			
+			try{
+				$html =
+					$this->get(
+						$proxy,
+						"https://search.marginalia.nu/search?" . $params
+					);
+			}catch(Exception $error){
 				
-				switch($v){
+				throw new Exception("Failed to get HTML");
+			}
+			
+		}else{
+			$params = [
+				"query" => $search
+			];
+			
+			foreach(["adtech", "recent", "intitle"] as $v){
+				
+				if($get[$v] == "yes"){
 					
-					case "adtech": $params["adtech"] = "reduce"; break;
-					case "recent": $params["recent"] = "recent"; break;
-					case "adtech": $params["searchTitle"] = "title"; break;
+					switch($v){
+						
+						case "adtech": $params["adtech"] = "reduce"; break;
+						case "recent": $params["recent"] = "recent"; break;
+						case "adtech": $params["searchTitle"] = "title"; break;
+					}
 				}
 			}
-		}
-		
-		try{
-			$html =
-				$this->get(
-					$this->backend->get_ip(),
-					"https://search.marginalia.nu/search",
-					$params
-				);
-		}catch(Exception $error){
 			
-			throw new Exception("Failed to get HTML");
+			try{
+				$html =
+					$this->get(
+						$proxy,
+						"https://search.marginalia.nu/search",
+						$params
+					);
+			}catch(Exception $error){
+				
+				throw new Exception("Failed to get HTML");
+			}
 		}
 		
 		$this->fuckhtml->load($html);
@@ -387,6 +411,65 @@ class marginalia{
 			];
 		}
 		
+		// get next page
+		$this->fuckhtml->load($html);
+		
+		$pagination =
+			$this->fuckhtml
+			->getElementsByAttributeValue(
+				"aria-label",
+				"pagination",
+				"nav"
+			);
+		
+		if(count($pagination) === 0){
+			
+			// no pagination
+			return $out;
+		}
+		
+		$this->fuckhtml->load($pagination[0]);
+		
+		$pages =
+			$this->fuckhtml
+			->getElementsByClassName(
+				"page-link",
+				"a"
+			);
+		
+		$found_current_page = false;
+		
+		foreach($pages as $page){
+			
+			if(
+				stripos(
+					$page["attributes"]["class"],
+					"active"
+				) !== false
+			){
+				
+				$found_current_page = true;
+				continue;
+			}
+			
+			if($found_current_page){
+				
+				// we found current page index, and we iterated over
+				// the next page <a>
+				
+				$out["npt"] =
+					$this->backend->store(
+						parse_url(
+							$page["attributes"]["href"],
+							PHP_URL_QUERY
+						),
+						"web",
+						$proxy
+					);
+				break;
+			}
+		}
+		
 		return $out;
 	}
 }

scraper/mojeek.php

@@ -701,9 +701,11 @@ class mojeek{
 				if(count($thumb) === 2){
 					
 					$answer["thumb"] =
-						$this->fuckhtml
-						->getTextContent(
-							$thumb[1]
+						urldecode(
+							$this->fuckhtml
+							->getTextContent(
+								$thumb[1]
+							)
 						);
 				}
 			}

settings.php

@@ -133,6 +133,10 @@ $settings = [
 						"value" => "google",
 						"text" => "Google"
 					],
+					[
+						"value" => "google_cse",
+						"text" => "Google CSE"
+					],
 					[
 						"value" => "startpage",
 						"text" => "Startpage"
@@ -203,6 +207,10 @@ $settings = [
 						"value" => "google",
 						"text" => "Google"
 					],
+					[
+						"value" => "google_cse",
+						"text" => "Google CSE"
+					],
 					[
 						"value" => "startpage",
 						"text" => "Startpage"

static/themes/Gore's shitty theme.css

@@ -16,6 +16,7 @@
 
 body{
 	padding:15px 4% 40px;
+	margin:unset;
 }
 
 h1,h2,h3,h4,h5,h6{

static/themes/Wine.css

@@ -0,0 +1,40 @@
+:root
+{
+        --accent : #f79e98;
+        --1d2021 : #180d0c;
+        --282828 : #180d0c;
+        --3c3836 : #251615;
+        --504945 : #251615;
+        --928374 : var(--accent);
+        --a89984 : #d8c5c4;
+        --bdae93 : #d8c5c4;
+        --8ec07c : var(--accent);
+        --ebdbb2 : #d8c5c4;
+        --comment: #928374;
+        --default: #DCC9BC;
+        --keyword: #F07342;
+        --string : var(--accent);
+        --green  : #959A6B;
+        --yellow : #E39C45;
+        --red    : #CF223E;
+        --white  : var(--a89984);
+        --black  : var(--1d2021);
+        --hover  : #b18884
+}
+
+      a.link, a { color: var(--accent); text-decoration: none; }
+      .searchbox { width: 23%; }
+      .filters filter select { color: #E39C45; }
+      .web .separator::before { color: var(--white) }
+      .searchbox input[type="text"]::placeholder { color: var(--white); }
+      a.link:hover
+      {
+        color: var(--hover);
+        text-shadow: 0 0 .2rem var(--hover);
+      }
+      .code-inline
+      { border-color: var(--default); font-family: monospace;}
+      .home #center a
+      { color: var(--accent); }
+      .home .subtext
+      { color: var(--white); }