add cnn selenium

.gitignore

@@ -3,7 +3,8 @@ composer.lock
 .env
 app/logs/*.log
 app/cache/*.gz
-app/cache/database/*.sql
+app/cache/database/.sqlite
+app/cache/*.json
 TODO.md
 node_modules
 

Dockerfile

@@ -10,6 +10,7 @@ RUN apt-get update && apt-get install -y \
     zip \
     git \
     htop \
+    cron \
     libzip-dev \
     libsqlite3-dev \
     && docker-php-ext-install zip opcache pdo_sqlite \
@@ -46,6 +47,8 @@ COPY default.conf /etc/nginx/sites-available/default
 # Copy and configure initialization script permissions
 COPY docker-entrypoint.sh /usr/local/bin/
 RUN chmod +x /usr/local/bin/docker-entrypoint.sh
+RUN chmod +x /app/bin/cleanup
+RUN chmod +x /app/bin/proxy
 
 # Create cache, database, and logs folders
 RUN mkdir -p /app/cache /app/cache/database /app/logs
@@ -54,6 +57,11 @@ RUN mkdir -p /app/cache /app/cache/database /app/logs
 RUN chown -R www-data:www-data /app \
     && chmod -R 755 /app
 
+# Configure Cron
+RUN touch /app/logs/cron.log
+RUN echo '0 * * * * root php "/app/bin/cleanup" >> /app/logs/cleanup.log 2>&1' >> /etc/crontab
+RUN echo '0 * * * * root php "/app/bin/proxy" >> /app/logs/proxy.log 2>&1' >> /etc/crontab
+
 EXPOSE 80
 
 ENTRYPOINT ["/usr/local/bin/docker-entrypoint.sh"]

README.en.md

@@ -27,6 +27,7 @@ Public instance at [marreta.pcdomanual.com](https://marreta.pcdomanual.com)!
 - Blocks domains you don't want
 - Allows configuring headers and cookies your way
 - PHP-FPM and OPcache
+- Proxy Support
 
 ## 🐳 Installing with Docker
 
@@ -65,6 +66,20 @@ Now just run `docker compose up -d`
 - S3 Cache: https://github.com/manualdousuario/marreta/wiki/%F0%9F%97%83%EF%B8%8F-Cache-S3
 - Maintenance: https://github.com/manualdousuario/marreta/wiki/%F0%9F%9B%A0%EF%B8%8F-Maintenance
 
+
+### 🛡️ DMCA
+
+To block domains from DMCA requests, create the file `app/cache/dmca_domains.json`:
+
+```json
+[
+    {
+        "host": "exemplo.com.br",
+        "message": "This content has been blocked on request"
+    }
+]
+```
+
 ## 🚀 Integrations
 
 - 🤖 **Telegram**: [Official Bot](https://t.me/leissoai_bot)

README.md

@@ -25,8 +25,10 @@ Instancia publica em [marreta.pcdomanual.com](https://marreta.pcdomanual.com)!
 - Remove elementos indesejados
 - Cache, cache!
 - Bloqueia domínios que você não quer
+- Proteção DMCA com mensagens personalizadas
 - Permite configurar headers e cookies do seu jeito
 - PHP-FPM e OPcache
+- Suporte a Proxy
 
 ## 🐳 Instalando em Docker
 
@@ -65,6 +67,19 @@ Agora só rodar `docker compose up -d`
 - Cache S3: https://github.com/manualdousuario/marreta/wiki/%F0%9F%97%83%EF%B8%8F-Cache-S3
 - Manutenção: https://github.com/manualdousuario/marreta/wiki/%F0%9F%9B%A0%EF%B8%8F-Maintenance
 
+### 🛡️ DMCA
+
+Para bloquear dominios por pedidos de DMCA, crie o arquivo `app/cache/dmca_domains.json`:
+
+```json
+[
+    {
+        "host": "exemplo.com.br",
+        "message": "Este conteúdo foi bloqueado a pedido"
+    }
+]
+```
+
 ## 🚀 Integrações
 
 - 🤖 **Telegram**: [Bot oficial](https://t.me/leissoai_bot)

app/.env.sample

@@ -42,3 +42,16 @@ SELENIUM_HOST=localhost:4444
 
 # Debug Settings
 DEBUG=false
+
+# Cache Cleanup Settings
+# Number of days to keep cache files (*.gz)
+# If not set, no files will be cleaned
+CLEANUP_DAYS=7
+
+# Proxy List Configuration
+# URL to download proxy list from (used by bin/proxy script)
+# The proxy list should contain proxies in one of these formats:
+# 1. http://USER:PASSWORD@HOST:PORT
+# 2. IP:PORT:USER:PASSWORD
+# Example: PROXY_LIST=https://example.com/proxy-list.txt
+PROXY_LIST=

app/assets/icons/moon.svg

@@ -0,0 +1,3 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" fill="currentColor" class="bi bi-moon-fill" viewBox="0 0 16 16">
+  <path d="M6 .278a.77.77 0 0 1 .08.858 7.2 7.2 0 0 0-.878 3.46c0 4.021 3.278 7.277 7.318 7.277q.792-.001 1.533-.16a.79.79 0 0 1 .81.316.73.73 0 0 1-.031.893A8.35 8.35 0 0 1 8.344 16C3.734 16 0 12.286 0 7.71 0 4.266 2.114 1.312 5.124.06A.75.75 0 0 1 6 .278"/>
+</svg>

app/assets/icons/sun.svg

@@ -0,0 +1,3 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" fill="currentColor" class="bi bi-brightness-high-fill" viewBox="0 0 16 16">
+  <path d="M12 8a4 4 0 1 1-8 0 4 4 0 0 1 8 0M8 0a.5.5 0 0 1 .5.5v2a.5.5 0 0 1-1 0v-2A.5.5 0 0 1 8 0m0 13a.5.5 0 0 1 .5.5v2a.5.5 0 0 1-1 0v-2A.5.5 0 0 1 8 13m8-5a.5.5 0 0 1-.5.5h-2a.5.5 0 0 1 0-1h2a.5.5 0 0 1 .5.5M3 8a.5.5 0 0 1-.5.5h-2a.5.5 0 0 1 0-1h2A.5.5 0 0 1 3 8m10.657-5.657a.5.5 0 0 1 0 .707l-1.414 1.415a.5.5 0 1 1-.707-.708l1.414-1.414a.5.5 0 0 1 .707 0m-9.193 9.193a.5.5 0 0 1 0 .707L3.05 13.657a.5.5 0 0 1-.707-.707l1.414-1.414a.5.5 0 0 1 .707 0m9.193 2.121a.5.5 0 0 1-.707 0l-1.414-1.414a.5.5 0 0 1 .707-.707l1.414 1.414a.5.5 0 0 1 0 .707M4.464 4.465a.5.5 0 0 1-.707 0L2.343 3.05a.5.5 0 1 1 .707-.707l1.414 1.414a.5.5 0 0 1 0 .708"/>
+</svg>

app/assets/js/scripts.js

@@ -101,4 +101,20 @@ document.addEventListener('DOMContentLoaded', function () {
             }
         });
     }
+
+    // Dark mode 
+    const themeToggle = document.getElementById('themeToggle');
+    const html = document.documentElement;
+    const savedTheme = localStorage.getItem('theme') || 'light';
+    html.setAttribute('data-theme', savedTheme);
+
+    if (themeToggle) {
+        themeToggle.addEventListener('click', () => {
+            const currentTheme = html.getAttribute('data-theme');
+            const newTheme = currentTheme === 'dark' ? 'light' : 'dark';
+            
+            html.setAttribute('data-theme', newTheme);
+            localStorage.setItem('theme', newTheme);
+        });
+    }
 });

app/assets/scss/_fonts.scss

@@ -1,10 +1,10 @@
 @font-face {
     font-family: 'inter';
-    src: url('../dist/fonts/inter-500.eot');
+    src: url('/dist/fonts/inter-500.eot');
     src: local('Inter Medium'), local('Inter-Medium'),
-        url('../dist/fonts/inter-500.woff2') format('woff2'),
-        url('../dist/fonts/inter-500.woff') format('woff'),
-        url('../dist/fonts/inter-500.ttf') format('truetype');
+        url('/dist/fonts/inter-500.woff2') format('woff2'),
+        url('/dist/fonts/inter-500.woff') format('woff'),
+        url('/dist/fonts/inter-500.ttf') format('truetype');
     font-weight: 500;
     font-style: normal;
     font-display: swap;
@@ -12,11 +12,11 @@
 
 @font-face {
     font-family: 'inter';
-    src: url('../dist/fonts/inter-600.eot');
+    src: url('/dist/fonts/inter-600.eot');
     src: local('Inter SemiBold'), local('Inter-SemiBold'),
-        url('../dist/fonts/inter-600.woff2') format('woff2'),
-        url('../dist/fonts/inter-600.woff') format('woff'),
-        url('../dist/fonts/inter-600.ttf') format('truetype');
+        url('/dist/fonts/inter-600.woff2') format('woff2'),
+        url('/dist/fonts/inter-600.woff') format('woff'),
+        url('/dist/fonts/inter-600.ttf') format('truetype');
     font-weight: 600;
     font-style: normal;
     font-display: swap;
@@ -24,11 +24,11 @@
 
 @font-face {
     font-family: 'unna';
-    src: url('../dist/fonts/unna-400.eot');
+    src: url('/dist/fonts/unna-400.eot');
     src: local('Unna Regular'), local('Unna-Regular'),
-        url('../dist/fonts/unna-400.woff2') format('woff2'),
-        url('../dist/fonts/unna-400.woff') format('woff'),
-        url('../dist/fonts/unna-400.ttf') format('truetype');
+        url('/dist/fonts/unna-400.woff2') format('woff2'),
+        url('/dist/fonts/unna-400.woff') format('woff'),
+        url('/dist/fonts/unna-400.ttf') format('truetype');
     font-weight: 400;
     font-style: normal;
     font-display: swap;

app/assets/scss/_icons.scss

@@ -41,3 +41,6 @@
 @include mixin.icon('close', 'invert(100%) sepia(32%) saturate(8%) hue-rotate(23deg) brightness(102%) contrast(100%)');
 
 @include mixin.icon('paste', 'invert(0%) sepia(21%) saturate(7425%) hue-rotate(12deg) brightness(96%) contrast(96%)');
+
+@include mixin.icon('sun', 'invert(0%) sepia(21%) saturate(7425%) hue-rotate(12deg) brightness(96%) contrast(96%)');
+@include mixin.icon('moon', 'invert(0%) sepia(21%) saturate(7425%) hue-rotate(12deg) brightness(96%) contrast(96%)');

app/assets/scss/_mixin.scss

@@ -17,7 +17,7 @@
 
 @mixin icon($name, $filter) {
 	.icon--#{$name} {
-		background-image: url("../dist/icons/#{$name}.svg");
+		background-image: url("/dist/icons/#{$name}.svg");
         filter: #{$filter};
 	}
 }

app/assets/scss/_root.scss

@@ -27,17 +27,53 @@
 	--font-weight: 500;
 	--line-height: 160%;
 
+	/* Light theme colors */
 	@include mixin.create-color('marreta', #3B82F6);
     @include mixin.create-color('text', #484848);
     @include mixin.create-color('textmuted', #818181);
     @include mixin.create-color('link', #3B82F6);
     
+    /* Theme-aware colors */
+    --background: #ffffff;
+    --surface: #F4F4F5;
+    --surface-hover: #e4e4e7;
+    --border: #e4e4e7;
+    --header-text: #000000;
+    --nav-mobile-bg: var(--marreta);
+    --nav-mobile-text: #ffffff;
+    --nav-desktop-text: #333333;
+    --nav-desktop-hover: #007bff;
+    --input-bg: #F4F4F5;
+    --toast-error: rgb(247, 102, 97);
+    --toast-warning: rgb(247, 152, 97);
+
 	--container_spacing: 24px;
 	@include mixin.devices(desktop) {
         --container_spacing: 64px;
     }
 }
 
+/* Dark theme */
+[data-theme="dark"] {
+    @include mixin.create-color('marreta', #60A5FA);
+    @include mixin.create-color('text', #e5e5e5);
+    @include mixin.create-color('textmuted', #a1a1aa);
+    @include mixin.create-color('link', #60A5FA);
+    
+    --background: #000;
+    --surface: #1f1f1f;
+    --surface-hover: #2a2a2a;
+    --border: #2a2a2a;
+    --header-text: #ffffff;
+    --nav-mobile-bg: var(--marreta);
+    --nav-mobile-text: #ffffff;
+    --nav-desktop-text: #e5e5e5;
+    --nav-desktop-hover: #60A5FA;
+    --input-bg: #1f1f1f;
+    --toast-error: rgb(220, 38, 127);
+    --toast-warning: rgb(245, 158, 11);
+}
+
 html {
 	scroll-behavior: smooth;
 }

app/assets/scss/home.scss

@@ -46,22 +46,22 @@ body {
     }
 
     &--error {
-        background-color: rgb(247, 102, 97);
+        background-color: var(--toast-error);
     }
 
     &--warning {
-        background-color: rgb(247, 152, 97);
+        background-color: var(--toast-warning);
     }
 }
 
 header {
     display: grid;
-    grid-template-columns: auto 1fr 1fr;
+    grid-template-columns: auto 1fr auto 1fr;
     align-items: center;
     padding: 0 0 42px 0;
 
     @include mixin.devices(desktop) {
-        grid-template-columns: 1fr 2fr 1fr;
+        grid-template-columns: 1fr 2fr auto 1fr;
     }
 
     &.open {
@@ -126,7 +126,77 @@ header {
 
         h1 {
             font-family: var(--font-family-unna);
-            color: #000;
+            color: var(--header-text);
+        }
+    }
+
+    .fast_buttons {
+        display: flex;
+        gap: 8px;
+    }
+
+    .theme-controls {
+        display: flex;
+        justify-content: center;
+        align-items: center;
+        padding: 0 16px;
+
+        @include mixin.devices(desktop) {
+            padding: 0;
+        }
+
+        .theme-toggle {
+            background: none;
+            border: 2px solid var(--border);
+            border-radius: 50%;
+            width: 40px;
+            height: 40px;
+            cursor: pointer;
+            display: flex;
+            align-items: center;
+            justify-content: center;
+            position: relative;
+            transition: all 0.3s ease;
+
+            &:hover {
+                border-color: var(--marreta);
+                background-color: var(--surface-hover);
+            }
+
+            .icon {
+                width: 18px;
+                height: 18px;
+                position: absolute;
+                transition: all 0.3s ease;
+                
+                &--sun {
+                    opacity: 1;
+                    transform: rotate(0deg) scale(1);
+                }
+                
+                &--moon {
+                    opacity: 0;
+                    transform: rotate(180deg) scale(0.8);
+                }
+                
+                [data-theme="dark"] & {
+                    filter: invert(1);
+                }
+            }
+
+            [data-theme="dark"] & {
+                .icon {
+                    &--sun {
+                        opacity: 0;
+                        transform: rotate(-180deg) scale(0.8);
+                    }
+                    
+                    &--moon {
+                        opacity: 1;
+                        transform: rotate(0deg) scale(1);
+                    }
+                }
+            }
         }
     }
 
@@ -137,7 +207,7 @@ header {
         left: 0;
         right: 0;
         bottom: 0;
-        background-color: var(--marreta);
+        background-color: var(--nav-mobile-bg);
         padding: var(--container_spacing) var(--container_spacing) calc(4*var(--container_spacing)) var(--container_spacing);
         z-index: 500;
         align-items: flex-end;
@@ -172,20 +242,20 @@ header {
             font-size: 24px;
             padding: 16px 0;
             border-bottom: 1px solid rgba(255, 255, 255, 0.24);
-            color: #fff;
+            color: var(--nav-mobile-text);
             text-decoration: none;
 
             @include mixin.devices(desktop) {
-                color: #333;
+                color: var(--nav-desktop-text);
                 font-size: initial;
                 padding: 0;
                 border-bottom: 0;
             }
 
             &:hover {
-                color: #fff;
+                color: var(--nav-mobile-text);
                 @include mixin.devices(desktop) {
-                    color: #007bff;
+                    color: var(--nav-desktop-hover);
                 }
             }
         }
@@ -204,14 +274,14 @@ header {
                 color: rgba(255,255,255,0.5);
                 padding: 0;
                 @include mixin.devices(desktop) {
-                    color: #333;
+                    color: var(--nav-desktop-text);
                     cursor: pointer;
                 }
 
                 &:hover {
                     color: rgba(255,255,255,0.5);
                     @include mixin.devices(desktop) {
-                        color: #007bff;
+                        color: var(--nav-desktop-hover);
                     }
                 }
             }
@@ -222,8 +292,8 @@ header {
                     top: 110%;
                     left: 0;
                     border-radius: 16px;
-                    background-color: #F4F4F5;
-                    border: 4px solid #F4F4F5;
+                    background-color: var(--surface);
+                    border: 4px solid var(--surface);
                     z-index: 10;
                     box-shadow: 0px 4px 6px 0px rgba(0, 0, 0, 0.05);
                     box-shadow: 0px 10px 15px 0px rgba(0, 0, 0, 0.1);
@@ -246,7 +316,7 @@ header {
                     font-weight: 600;
 
                     @include mixin.devices(desktop) {
-                        background-color: #fff;
+                        background-color: var(--background);
                         margin-top: 0;
                         margin-bottom: 4px;
                         padding: 8px 16px;
@@ -358,8 +428,8 @@ header {
             top: 110%;
             right: 0;
             border-radius: 16px;
-            background-color: #F4F4F5;
-            border: 4px solid #F4F4F5;
+            background-color: var(--surface);
+            border: 4px solid var(--surface);
             z-index: 10;
             box-shadow: 0px 4px 6px 0px rgba(0, 0, 0, 0.05);
             box-shadow: 0px 10px 15px 0px rgba(0, 0, 0, 0.1);
@@ -388,7 +458,7 @@ header {
                 font-weight: 600;
                 display: block;
                 padding: 8px 16px;
-                background-color: #fff;
+                background-color: var(--background);
                 display: flex;
                 align-items: center;
 
@@ -408,7 +478,7 @@ header {
 
         &.open {
             .extension__toggle {
-                background-color: #F4F4F5;
+                background-color: var(--surface);
                 color: var(--textmuted);
             }
 
@@ -428,7 +498,7 @@ main {
         font-size: 64px;
         line-height: 61.44px;
         text-align: center;
-        color: #000;
+        color: var(--header-text);
         max-width: 512px;
         margin: 0 auto;
     }
@@ -452,7 +522,7 @@ main {
         .fields {
             &::before {
                 content: '';
-                background-image: url(../assets/images/wall.png);
+                background-image: url(/assets/images/wall.png);
                 background-repeat: no-repeat;
                 background-size: 100% 100%;
                 width: 422px;
@@ -461,6 +531,11 @@ main {
                 top: -110px;
                 right: -180px;
                 z-index: 1;
+                transition: filter 0.3s ease;
+                
+                [data-theme="dark"] & {
+                    filter: invert(1);
+                }
             }
 
             max-width: 470px;
@@ -485,7 +560,7 @@ main {
                 }
 
                 input {
-                    background-color: #F4F4F5;
+                    background-color: var(--input-bg);
                     padding: 16px 0 16px 44px;
                     border: 0;
                     border-radius: 8px;
@@ -493,12 +568,13 @@ main {
                     box-sizing: border-box;
                     position: relative;
                     line-height: 1.3em;
+                    color: var(--text);
                 }
             }
 
             .paste {
-                background: rgb(244,244,245);
-                background: linear-gradient(90deg, rgba(244,244,245,0) 0%, rgba(244,244,245,1) 30%, rgba(244,244,245,1) 100%);
+                background: var(--input-bg);
+                background: linear-gradient(90deg, transparent 0%, var(--input-bg) 30%, var(--input-bg) 100%);
                 align-items: center;
                 z-index: 3;
                 position: absolute;
@@ -508,8 +584,13 @@ main {
                 cursor: pointer;
                 height: 48px;
                 display: flex;
-                .icon {
                 
+                .icon {
+                    transition: filter 0.3s ease;
+                    
+                    [data-theme="dark"] & {
+                        filter: invert(1);
+                    }
                 }
             }
 
@@ -557,7 +638,7 @@ main {
     .plus {
         z-index: 3;
         position: relative;
-        background-color: rgba(244, 244, 245, 1);
+        background-color: var(--surface);
         margin-left: calc(-1*var(--container_spacing));
         margin-right: calc(-1*var(--container_spacing));
 

app/bin/cleanup

@@ -0,0 +1,211 @@
+#!/usr/bin/env php
+<?php
+
+/**
+ * Cache Cleanup Script
+ *
+ * Removes *.gz files from the cache directory that are older than the number
+ * of days specified in the CLEANUP_DAYS environment variable.
+ * If CLEANUP_DAYS is not set, no files will be cleaned.
+ */
+
+require_once __DIR__ . '/../vendor/autoload.php';
+
+use League\CLImate\CLImate;
+use Dotenv\Dotenv;
+use Aws\S3\S3Client;
+use Aws\Exception\AwsException;
+
+$climate = new CLImate();
+$climate->bold()->out('Cache Cleanup Tool');
+$climate->br();
+
+$cleanupDays = 0;
+
+try {
+    $dotenv = Dotenv::createImmutable(__DIR__ . '/..');
+    $dotenv->load();
+    $climate->out('Environment variables loaded');
+    $cleanupDays = $_ENV['CLEANUP_DAYS'];
+} catch (\Exception $e) {
+    $climate->yellow()->out('Warning: ' . $e->getMessage());
+    exit(0);
+}
+
+if (!defined('CACHE_DIR')) {
+    define('CACHE_DIR', __DIR__ . '/../cache');
+}
+
+if ($cleanupDays == 0) {
+    $climate->yellow()->out('CLEANUP_DAYS variable not set or 0. No files will be cleaned.');
+    exit(0);
+}
+
+$cleanupDays = (int)$cleanupDays;
+if ($cleanupDays <= 0) {
+    $climate->red()->out('CLEANUP_DAYS must be a positive integer. No files will be cleaned.');
+    exit(1);
+};
+
+// Calculate the cutoff timestamp
+$cutoffTime = time() - ($cleanupDays * 86400);
+
+// Check if S3 cache is enabled
+$s3CacheEnabled = isset($_ENV['S3_CACHE_ENABLED']) && filter_var($_ENV['S3_CACHE_ENABLED'], FILTER_VALIDATE_BOOLEAN);
+
+if ($s3CacheEnabled) {
+    // Clean S3 cache
+    cleanS3Cache($climate, $cutoffTime, $cleanupDays);
+} else {
+    // Clean local disk cache
+    cleanDiskCache($climate, $cutoffTime, $cleanupDays);
+}
+
+/**
+ * Clean cache files from S3 bucket
+ * 
+ * @param CLImate $climate CLImate instance for output
+ * @param int $cutoffTime Timestamp to use as cutoff for file age
+ * @param int $cleanupDays Number of days to keep files
+ */
+function cleanS3Cache($climate, $cutoffTime, $cleanupDays) {
+    $requiredVars = ['S3_ACCESS_KEY', 'S3_SECRET_KEY', 'S3_BUCKET'];
+    foreach ($requiredVars as $var) {
+        if (!isset($_ENV[$var]) || empty($_ENV[$var])) {
+            $climate->red()->out("$var environment variable is required for S3 cache cleaning.");
+            exit(1);
+        }
+    }
+
+    $climate->out("S3 cache enabled. Cleaning S3 cache files older than {$cleanupDays} days...");
+
+    $clientConfig = [
+        'version' => 'latest',
+        'region'  => $_ENV['S3_REGION'] ?? 'us-east-1',
+        'credentials' => [
+            'key'    => $_ENV['S3_ACCESS_KEY'],
+            'secret' => $_ENV['S3_SECRET_KEY'],
+        ]
+    ];
+
+    if (!empty($_ENV['S3_ENDPOINT'])) {
+        $clientConfig['endpoint'] = $_ENV['S3_ENDPOINT'];
+        $clientConfig['use_path_style_endpoint'] = true;
+    }
+
+    try {
+        $s3Client = new S3Client($clientConfig);
+        $bucket = $_ENV['S3_BUCKET'];
+        $prefix = $_ENV['S3_FOLDER'] ?? 'cache/';
+
+        $climate->out("Listing objects in bucket: {$bucket} with prefix: {$prefix}");
+        
+        $objects = [];
+        $marker = null;
+        
+        do {
+            $params = [
+                'Bucket' => $bucket,
+                'Prefix' => $prefix,
+                'MaxKeys' => 1000
+            ];
+            
+            if ($marker) {
+                $params['Marker'] = $marker;
+            }
+            
+            $result = $s3Client->listObjects($params);
+            
+            if (isset($result['Contents'])) {
+                foreach ($result['Contents'] as $object) {
+                    if (substr($object['Key'], -3) === '.gz') {
+                        $objects[] = $object;
+                    }
+                }
+            }
+            
+            $marker = $result['NextMarker'] ?? ($result['IsTruncated'] ? end($result['Contents'])['Key'] : null);
+        } while ($marker);
+
+        $totalObjects = count($objects);
+        $climate->out("Found {$totalObjects} .gz objects in S3 bucket.");
+
+        if ($totalObjects === 0) {
+            $climate->out('No .gz objects found in S3 bucket.');
+            return;
+        }
+
+        $progress = $climate->progress()->total($totalObjects);
+        $deletedObjects = 0;
+
+        foreach ($objects as $index => $object) {
+            $progress->current($index + 1);
+            $lastModified = strtotime($object['LastModified']);
+
+            if ($lastModified < $cutoffTime) {
+                try {
+                    $s3Client->deleteObject([
+                        'Bucket' => $bucket,
+                        'Key'    => $object['Key']
+                    ]);
+                    $deletedObjects++;
+                } catch (AwsException $e) {
+                    $climate->red()->out("Failed to delete: " . $object['Key'] . " - " . $e->getMessage());
+                }
+            }
+        }
+
+        $climate->br();
+        $climate->green()->out("S3 cleanup complete: {$deletedObjects} objects deleted.");
+    } catch (AwsException $e) {
+        $climate->red()->out("AWS Error: " . $e->getMessage());
+        exit(1);
+    }
+}
+
+/**
+ * Clean cache files from local disk
+ * 
+ * @param CLImate $climate CLImate instance for output
+ * @param int $cutoffTime Timestamp to use as cutoff for file age
+ * @param int $cleanupDays Number of days to keep files
+ */
+function cleanDiskCache($climate, $cutoffTime, $cleanupDays) {
+    $cacheDir = CACHE_DIR;
+
+    $climate->out("Cleaning cache files older than {$cleanupDays} days from: {$cacheDir}");
+
+    if (!is_dir($cacheDir)) {
+        $climate->red()->out("Cache directory not found: {$cacheDir}");
+        exit(1);
+    }
+
+    $gzFiles = glob($cacheDir . '/*.gz');
+    $totalFiles = count($gzFiles);
+    $deletedFiles = 0;
+
+    if ($totalFiles === 0) {
+        $climate->out('No .gz files found in cache directory.');
+        return;
+    }
+
+    $climate->out("Found {$totalFiles} .gz files in cache directory.");
+
+    $progress = $climate->progress()->total($totalFiles);
+
+    foreach ($gzFiles as $index => $file) {
+        $progress->current($index + 1);
+        $fileTime = filemtime($file);
+
+        if ($fileTime < $cutoffTime) {
+            if (unlink($file)) {
+                $deletedFiles++;
+            } else {
+                $climate->red()->out("Failed to delete: " . basename($file));
+            }
+        }
+    }
+
+    $climate->br();
+    $climate->green()->out("Disk cleanup complete: {$deletedFiles} files deleted.");
+}

app/bin/proxy

@@ -0,0 +1,196 @@
+#!/usr/bin/env php
+<?php
+
+/**
+ * Proxy List Cache Updater
+ *
+ * Downloads proxy list from the URL specified in the PROXY_LIST environment variable
+ * and stores it in the cache directory for reuse.
+ * This script should be run daily via cron to keep the proxy list updated.
+ *
+ * Supported proxy list formats:
+ * 1. http://USER:PASSWORD@HOST:PORT
+ * 2. IP:PORT:USER:PASSWORD
+ */
+
+require_once __DIR__ . '/../vendor/autoload.php';
+
+use League\CLImate\CLImate;
+use Dotenv\Dotenv;
+use Curl\Curl;
+
+$climate = new CLImate();
+$climate->bold()->out('Proxy List Cache Updater');
+$climate->br();
+
+try {
+    $dotenv = Dotenv::createImmutable(__DIR__ . '/..');
+    $dotenv->load();
+    $climate->out('Environment variables loaded');
+} catch (\Exception $e) {
+    $climate->yellow()->out('Warning: ' . $e->getMessage());
+    exit(0);
+}
+
+if (!defined('CACHE_DIR')) {
+    define('CACHE_DIR', __DIR__ . '/../cache');
+}
+
+if (!isset($_ENV['PROXY_LIST']) || empty($_ENV['PROXY_LIST'])) {
+    $climate->yellow()->out('PROXY_LIST environment variable not set. No proxies to cache.');
+    exit(0);
+}
+
+$proxyListUrl = $_ENV['PROXY_LIST'];
+$proxyCachePath = CACHE_DIR . '/proxy_list.json';
+
+// Download proxy list from URL
+$climate->out('Downloading proxy list from: ' . $proxyListUrl);
+$proxyList = downloadProxyList($proxyListUrl, $climate);
+
+if ($proxyList === false) {
+    $climate->red()->out('Failed to download proxy list from URL: ' . $proxyListUrl);
+    exit(1);
+}
+
+$climate->green()->out('Proxy list downloaded successfully (' . strlen($proxyList) . ' bytes)');
+
+if (!is_dir(CACHE_DIR)) {
+    if (!mkdir(CACHE_DIR, 0755, true)) {
+        $climate->red()->out('Failed to create cache directory: ' . CACHE_DIR);
+        exit(1);
+    }
+}
+
+$climate->out('Parsing proxy list from environment variable...');
+$proxies = parseProxyList($proxyList);
+
+if (empty($proxies)) {
+    $climate->red()->out('No valid proxies found in PROXY_LIST. Supported formats are:');
+    $climate->red()->out('1. http://USER:PASSWORD@HOST:PORT');
+    $climate->red()->out('2. IP:PORT:USER:PASSWORD');
+    exit(1);
+}
+
+$climate->out('Found ' . count($proxies) . ' valid proxies.');
+
+if (file_put_contents($proxyCachePath, json_encode($proxies))) {
+    $climate->green()->out('Proxy list successfully cached to: ' . $proxyCachePath);
+} else {
+    $climate->red()->out('Failed to write proxy list to cache file: ' . $proxyCachePath);
+    exit(1);
+}
+
+/**
+ * Parse proxy list from environment variable
+ *
+ * @param string $proxyListString Proxy list in format http://USER:PASSWORD@HOST:PORT or IP:PORT:USER:PASSWORD
+ * @return array Array of valid proxy URLs
+ */
+function parseProxyList($proxyListString) {
+    $proxies = [];
+    $lines = preg_split('/[\r\n,]+/', $proxyListString);
+    
+    foreach ($lines as $line) {
+        $line = trim($line);
+        if (empty($line)) continue;
+        
+        // Format 1: http://USER:PASSWORD@HOST:PORT
+        if (preg_match('/^https?:\/\/[^:]+:[^@]+@[^:]+:\d+$/i', $line)) {
+            $proxies[] = $line;
+            continue;
+        }
+        
+        // Format 2: IP:PORT:USER:PASSWORD
+        if (preg_match('/^(\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}):(\d+):([^:]+):(.+)$/', $line, $matches)) {
+            $ip = $matches[1];
+            $port = $matches[2];
+            $user = $matches[3];
+            $password = $matches[4];
+            
+            // Convert to standard format
+            $proxies[] = "http://{$user}:{$password}@{$ip}:{$port}";
+        }
+    }
+    
+    return $proxies;
+}
+
+/**
+ * Download proxy list from URL using php-curl-class
+ *
+ * @param string $url URL to download proxy list from
+ * @param CLImate $climate CLImate instance for output
+ * @return string|false Downloaded content or false on failure
+ */
+function downloadProxyList($url, $climate = null) {
+    $curl = new Curl();
+    
+    // Configure cURL options
+    $curl->setTimeout(30);
+    $curl->setConnectTimeout(10);
+    $curl->setUserAgent('Marreta Proxy Updater/1.0');
+    $curl->setHeader('Accept', 'text/plain, text/html, */*');
+    $curl->setHeader('Accept-Encoding', 'gzip, deflate');
+    $curl->setOpt(CURLOPT_FOLLOWLOCATION, true);
+    $curl->setOpt(CURLOPT_MAXREDIRS, 3);
+    $curl->setOpt(CURLOPT_SSL_VERIFYPEER, false);
+    $curl->setOpt(CURLOPT_SSL_VERIFYHOST, false);
+    
+    try {
+        if ($climate) {
+            $climate->out('Making HTTP request with php-curl-class...');
+        }
+        
+        $curl->get($url);
+        
+        if ($curl->error) {
+            $errorMsg = 'cURL request failed: ' . $curl->errorMessage . ' (Code: ' . $curl->errorCode . ')';
+            
+            if ($climate) {
+                $climate->red()->out($errorMsg);
+            } else {
+                error_log($errorMsg);
+            }
+            
+            return false;
+        }
+        
+        $statusCode = $curl->httpStatusCode;
+        
+        if ($climate) {
+            $climate->out('HTTP Status Code: ' . $statusCode);
+        }
+        
+        if ($statusCode === 200) {
+            $content = $curl->response;
+            
+            if ($climate) {
+                $contentType = $curl->responseHeaders['Content-Type'] ?? 'unknown';
+                $climate->out('Content-Type: ' . $contentType);
+                $climate->out('Content-Length: ' . strlen($content) . ' bytes');
+            }
+            
+            return $content;
+        }
+        
+        if ($climate) {
+            $climate->yellow()->out('Unexpected HTTP status code: ' . $statusCode);
+        }
+        
+        return false;
+        
+    } catch (\Exception $e) {
+        $errorMsg = 'Unexpected error during download: ' . $e->getMessage();
+        
+        if ($climate) {
+            $climate->red()->out($errorMsg);
+        } else {
+            error_log($errorMsg);
+        }
+        
+        return false;
+    } finally {
+        $curl->close();
+    }
+}

app/composer.json

@@ -5,7 +5,8 @@
         "php-curl-class/php-curl-class": "^11.0",
         "php-webdriver/webdriver": "^1.15",
         "monolog/monolog": "^3.8.1",
-        "nikic/fast-route": "^1.3"
+        "nikic/fast-route": "^1.3",
+        "league/climate": "^3.8"
     },
     "autoload": {
         "psr-4": {

app/config.php

@@ -21,15 +21,11 @@ try {
         'SITE_URL'
     ])->notEmpty();
 
-    // Validate URL format
-    if (!filter_var($_ENV['SITE_URL'], FILTER_VALIDATE_URL)) {
-        throw new Exception('SITE_URL must be a valid URL');
-    }
-
     // Core system settings
     define('SITE_NAME', $_ENV['SITE_NAME']);
     define('SITE_DESCRIPTION', $_ENV['SITE_DESCRIPTION']);
     define('SITE_URL', $_ENV['SITE_URL']);
+    define('CLEANUP_DAYS', $_ENV['CLEANUP_DAYS'] ?? 0);
     
     // Optional settings with defaults
     define('DNS_SERVERS', $_ENV['DNS_SERVERS'] ?? '1.1.1.1, 8.8.8.8');
@@ -66,6 +62,16 @@ try {
     define('DOMAIN_RULES', require __DIR__ . '/data/domain_rules.php');
     define('GLOBAL_RULES', require __DIR__ . '/data/global_rules.php');
     
+    // Load DMCA domains from JSON file
+    $dmcaDomainsFile = __DIR__ . '/cache/dmca_domains.json';
+    if (file_exists($dmcaDomainsFile)) {
+        $dmcaDomainsJson = file_get_contents($dmcaDomainsFile);
+        $dmcaDomains = json_decode($dmcaDomainsJson, true);
+        define('DMCA_DOMAINS', is_array($dmcaDomains) ? $dmcaDomains : []);
+    } else {
+        define('DMCA_DOMAINS', []);
+    }
+
 } catch (Dotenv\Exception\ValidationException $e) {
     die('Environment Error: ' . $e->getMessage());
 } catch (Exception $e) {

app/data/blocked_domains.php

@@ -6,7 +6,12 @@
  * Defines domains that cannot be accessed by the system
  * due to usage policies or technical restrictions
  */
+
+$host = parse_url(defined('SITE_URL') ? SITE_URL : '', PHP_URL_HOST);
 return [
+    $host,
+    'localhost',
+    '127.0.0.1',
     // News sites
     //-- Content behind login access/hard paywall
     'wsj.com',
@@ -121,7 +126,6 @@ return [
     'jusbrasil.com.br',
     'glassdoor.com.br',
     'gov.br',
-    'medium.com',
     'stackoverflow.com',
     'hoteis.com',
     'amazon.com',

app/data/domain_rules.php

@@ -13,6 +13,7 @@
  * - classAttrRemove: Array of classes to be removed from elements
  * - customCode: String containing custom JavaScript code
  * - customStyle: String containing custom CSS code
+ * - proxy: Enable proxy in Guzzle or Selenium requests
  * - excludeGlobalRules: Associative array of global rules to exclude for this domain
  *   Example:
  *   'excludeGlobalRules' => [
@@ -27,6 +28,16 @@
  * - fromGoogleBot: Adds simulation of request coming from Google Bot
  * - removeElementsByTag: Remove specific elements via DOM
  * - removeCustomAttr: Remove custom attributes from elements
+ * - urlMods: Modify the URL before fetching content.
+ *     Example:
+ *     'urlMods' => [
+ *         'query' => [
+ *             [
+ *                 'key' => 'amp',
+ *                 'value' => '1'
+ *             ]
+ *         ]
+ *     ]
  */
 return [
     'nsctotal.com.br' => [
@@ -40,6 +51,12 @@ return [
         'removeElementsByTag' => ['style'],
         'removeCustomAttr' => ['hidden', 'data-*']
     ],
+    'wired.com' => [
+        'scriptTagRemove' => ['.js'],
+    ],
+    'newyorker.com' => [
+        'scriptTagRemove' => ['.js'],
+    ],
     'globo.com' => [
         'idElementRemove' => ['cookie-banner-lgpd', 'paywall-cpt', 'mc-read-more-wrapper', 'paywall-cookie-content', 'paywall-cpt'],
         'classElementRemove' => ['banner-lgpd', 'article-related-link__title', 'article-related-link__picture', 'paywall-denied', 'banner-subscription'],
@@ -54,16 +71,19 @@ return [
     'gauchazh.clicrbs.com.br' => [
         'idElementRemove' => ['paywallTemplate'],
         'classAttrRemove' => ['m-paid-content', 'paid-content-apply'],
-        'scriptTagRemove' => ['vendors-8'],
+        'scriptTagRemove' => ['vendors-9','vendors-10','vendors-11'],
         'excludeGlobalRules' => [
             'classElementRemove' => ['paid-content']
         ],
-        'fetchStrategies' => 'fetchFromSelenium',
+        'proxy' => true,
     ],
     'reuters.com' => [
         'classElementRemove' => ['leaderboard__container'],
         'fetchStrategies' => 'fetchFromSelenium',
     ],
+    'cnn.com' => [
+        'fetchStrategies' => 'fetchFromSelenium',
+    ],
     'lepoint.fr' => [
         'classElementRemove' => ['paywall'],
     ],
@@ -86,10 +106,6 @@ return [
         'idElementRemove' => ['cboxOverlay'],
         'fetchStrategies' => 'fetchFromSelenium',
     ],
-    'washingtonpost.com' => [
-        'classElementRemove' => ['paywall-overlay'],
-        'fetchStrategies' => 'fetchFromSelenium',
-    ],
     'oantagonista.com.br' => [
         'fetchStrategies' => 'fetchFromSelenium',
     ],
@@ -170,6 +186,12 @@ return [
     'nzherald.co.nz' => [
         'fetchStrategies' => 'fetchFromSelenium',
     ],
+    'onetz.de' => [
+        'idElementRemove' => ['checkout-container'],
+        'classElementRemove' => ['tp-backdrop','dm-nobg'],
+        'classAttrRemove' => ['field-dnt-body-pp'],
+        'scriptTagRemove' => ['.js'],
+    ],
     'opovo.com.br' => [
         'classElementRemove' => ['screen-loading', 'overlay-advise']
     ],
@@ -199,13 +221,6 @@ return [
             }
         '
     ],
-    'ft.com' => [
-        'cookies' => [
-            'next-flags' => null,
-            'next:ads' => null
-        ],
-        'fromGoogleBot' => true
-    ],
     'nytimes.com' => [
         'idElementRemove' => ['gateway-content', 'site-index', 'complianceOverlay'],
         'customCode' => '
@@ -256,7 +271,7 @@ return [
                 position: relative !important;
             }
         ',
-        'fetchStrategies' => 'fetchFromSelenium',
+        'fetchStrategies' => 'fetchFromWaybackMachine',
         'excludeGlobalRules' => [
             'scriptTagRemove' => [
                 'gtm.js',
@@ -357,8 +372,358 @@ return [
             '_pctx' => null
         ]
     ],
-
-    // Domain test
+    'thestar.com' => [
+        'classElementRemove' => ['subscriber-offers', 'subscriber-only', 'subscription-required', 'redacted-overlay', 'subscriber-hide', 'tnt-ads-container'],
+        'customCode' => '
+            window.localStorage.clear();
+            document.addEventListener("DOMContentLoaded", () => {
+                const paywall = document.querySelectorAll(\'div.subscriber-offers\');
+                paywall.forEach(el => { el.remove(); });
+                const subscriber_only = document.querySelectorAll(\'div.subscriber-only\');
+                for (const elem of subscriber_only) {
+                    if (elem.classList.contains(\'encrypted-content\') && typeof DOMPurify !== \'undefined\' && typeof unscramble !== \'undefined\') {
+                        const parser = new DOMParser();
+                        const doc = parser.parseFromString(\'<div>\' + DOMPurify.sanitize(unscramble(elem.innerText)) + \'</div>\', \'text/html\');
+                        const content_new = doc.querySelector(\'div\');
+                        elem.parentNode.replaceChild(content_new, elem);
+                    }
+                    elem.removeAttribute(\'style\');
+                    elem.removeAttribute(\'class\');
+                }
+                const banners = document.querySelectorAll(\'div.subscription-required, div.redacted-overlay, div.subscriber-hide, div.tnt-ads-container\');
+                banners.forEach(el => { el.remove(); });
+                const ads = document.querySelectorAll(\'div.tnt-ads-container, div[class*="adLabelWrapper"]\');
+                ads.forEach(el => { el.remove(); });
+                const recommendations = document.querySelectorAll(\'div[id^="tncms-region-article"]\');
+                recommendations.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'niagarafallsreview.ca' => [
+        'classElementRemove' => ['subscriber-offers', 'subscriber-only', 'subscription-required', 'redacted-overlay', 'subscriber-hide', 'tnt-ads-container'],
+        'customCode' => '
+            window.localStorage.clear();
+            document.addEventListener("DOMContentLoaded", () => {
+                const paywall = document.querySelectorAll(\'div.subscriber-offers\');
+                paywall.forEach(el => { el.remove(); });
+                const subscriber_only = document.querySelectorAll(\'div.subscriber-only\');
+                for (const elem of subscriber_only) {
+                    if (elem.classList.contains(\'encrypted-content\') && typeof DOMPurify !== \'undefined\' && typeof unscramble !== \'undefined\') {
+                        const parser = new DOMParser();
+                        const doc = parser.parseFromString(\'<div>\' + DOMPurify.sanitize(unscramble(elem.innerText)) + \'</div>\', \'text/html\');
+                        const content_new = doc.querySelector(\'div\');
+                        elem.parentNode.replaceChild(content_new, elem);
+                    }
+                    elem.removeAttribute(\'style\');
+                    elem.removeAttribute(\'class\');
+                }
+                const banners = document.querySelectorAll(\'div.subscription-required, div.redacted-overlay, div.subscriber-hide, div.tnt-ads-container\');
+                banners.forEach(el => { el.remove(); });
+                const ads = document.querySelectorAll(\'div.tnt-ads-container, div[class*="adLabelWrapper"]\');
+                ads.forEach(el => { el.remove(); });
+                const recommendations = document.querySelectorAll(\'div[id^="tncms-region-article"]\');
+                recommendations.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'thepeterboroughexaminer.com' => [
+        'classElementRemove' => ['subscriber-offers', 'subscriber-only', 'subscription-required', 'redacted-overlay', 'subscriber-hide', 'tnt-ads-container'],
+        'customCode' => '
+            window.localStorage.clear();
+            document.addEventListener("DOMContentLoaded", () => {
+                const paywall = document.querySelectorAll(\'div.subscriber-offers\');
+                paywall.forEach(el => { el.remove(); });
+                const subscriber_only = document.querySelectorAll(\'div.subscriber-only\');
+                for (const elem of subscriber_only) {
+                    if (elem.classList.contains(\'encrypted-content\') && typeof DOMPurify !== \'undefined\' && typeof unscramble !== \'undefined\') {
+                        const parser = new DOMParser();
+                        const doc = parser.parseFromString(\'<div>\' + DOMPurify.sanitize(unscramble(elem.innerText)) + \'</div>\', \'text/html\');
+                        const content_new = doc.querySelector(\'div\');
+                        elem.parentNode.replaceChild(content_new, elem);
+                    }
+                    elem.removeAttribute(\'style\');
+                    elem.removeAttribute(\'class\');
+                }
+                const banners = document.querySelectorAll(\'div.subscription-required, div.redacted-overlay, div.subscriber-hide, div.tnt-ads-container\');
+                banners.forEach(el => { el.remove(); });
+                const ads = document.querySelectorAll(\'div.tnt-ads-container, div[class*="adLabelWrapper"]\');
+                ads.forEach(el => { el.remove(); });
+                const recommendations = document.querySelectorAll(\'div[id^="tncms-region-article"]\');
+                recommendations.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'therecord.com' => [
+        'classElementRemove' => ['subscriber-offers', 'subscriber-only', 'subscription-required', 'redacted-overlay', 'subscriber-hide', 'tnt-ads-container'],
+        'customCode' => '
+            window.localStorage.clear();
+            document.addEventListener("DOMContentLoaded", () => {
+                const paywall = document.querySelectorAll(\'div.subscriber-offers\');
+                paywall.forEach(el => { el.remove(); });
+                const subscriber_only = document.querySelectorAll(\'div.subscriber-only\');
+                for (const elem of subscriber_only) {
+                    if (elem.classList.contains(\'encrypted-content\') && typeof DOMPurify !== \'undefined\' && typeof unscramble !== \'undefined\') {
+                        const parser = new DOMParser();
+                        const doc = parser.parseFromString(\'<div>\' + DOMPurify.sanitize(unscramble(elem.innerText)) + \'</div>\', \'text/html\');
+                        const content_new = doc.querySelector(\'div\');
+                        elem.parentNode.replaceChild(content_new, elem);
+                    }
+                    elem.removeAttribute(\'style\');
+                    elem.removeAttribute(\'class\');
+                }
+                const banners = document.querySelectorAll(\'div.subscription-required, div.redacted-overlay, div.subscriber-hide, div.tnt-ads-container\');
+                banners.forEach(el => { el.remove(); });
+                const ads = document.querySelectorAll(\'div.tnt-ads-container, div[class*="adLabelWrapper"]\');
+                ads.forEach(el => { el.remove(); });
+                const recommendations = document.querySelectorAll(\'div[id^="tncms-region-article"]\');
+                recommendations.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'thespec.com' => [
+        'classElementRemove' => ['subscriber-offers', 'subscriber-only', 'subscription-required', 'redacted-overlay', 'subscriber-hide', 'tnt-ads-container'],
+        'customCode' => '
+            window.localStorage.clear();
+            document.addEventListener("DOMContentLoaded", () => {
+                const paywall = document.querySelectorAll(\'div.subscriber-offers\');
+                paywall.forEach(el => { el.remove(); });
+                const subscriber_only = document.querySelectorAll(\'div.subscriber-only\');
+                for (const elem of subscriber_only) {
+                    if (elem.classList.contains(\'encrypted-content\') && typeof DOMPurify !== \'undefined\' && typeof unscramble !== \'undefined\') {
+                        const parser = new DOMParser();
+                        const doc = parser.parseFromString(\'<div>\' + DOMPurify.sanitize(unscramble(elem.innerText)) + \'</div>\', \'text/html\');
+                        const content_new = doc.querySelector(\'div\');
+                        elem.parentNode.replaceChild(content_new, elem);
+                    }
+                    elem.removeAttribute(\'style\');
+                    elem.removeAttribute(\'class\');
+                }
+                const banners = document.querySelectorAll(\'div.subscription-required, div.redacted-overlay, div.subscriber-hide, div.tnt-ads-container\');
+                banners.forEach(el => { el.remove(); });
+                const ads = document.querySelectorAll(\'div.tnt-ads-container, div[class*="adLabelWrapper"]\');
+                ads.forEach(el => { el.remove(); });
+                const recommendations = document.querySelectorAll(\'div[id^="tncms-region-article"]\');
+                recommendations.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'wellandtribune.ca' => [
+        'classElementRemove' => ['subscriber-offers', 'subscriber-only', 'subscription-required', 'redacted-overlay', 'subscriber-hide', 'tnt-ads-container'],
+        'customCode' => '
+            window.localStorage.clear();
+            document.addEventListener("DOMContentLoaded", () => {
+                const paywall = document.querySelectorAll(\'div.subscriber-offers\');
+                paywall.forEach(el => { el.remove(); });
+                const subscriber_only = document.querySelectorAll(\'div.subscriber-only\');
+                for (const elem of subscriber_only) {
+                    if (elem.classList.contains(\'encrypted-content\') && typeof DOMPurify !== \'undefined\' && typeof unscramble !== \'undefined\') {
+                        const parser = new DOMParser();
+                        const doc = parser.parseFromString(\'<div>\' + DOMPurify.sanitize(unscramble(elem.innerText)) + \'</div>\', \'text/html\');
+                        const content_new = doc.querySelector(\'div\');
+                        elem.parentNode.replaceChild(content_new, elem);
+                    }
+                    elem.removeAttribute(\'style\');
+                    elem.removeAttribute(\'class\');
+                }
+                const banners = document.querySelectorAll(\'div.subscription-required, div.redacted-overlay, div.subscriber-hide, div.tnt-ads-container\');
+                banners.forEach(el => { el.remove(); });
+                const ads = document.querySelectorAll(\'div.tnt-ads-container, div[class*="adLabelWrapper"]\');
+                ads.forEach(el => { el.remove(); });
+                const recommendations = document.querySelectorAll(\'div[id^="tncms-region-article"]\');
+                recommendations.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'time.com' => [
+        'headers' => [
+            'User-Agent' => 'Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)',
+            'Cookie' => 'nyt-a=; nyt-gdpr=0; nyt-geo=DE; nyt-privacy=1',
+            'Referer' => 'https://www.google.com/'
+        ],
+        'customCode' => '
+            window.localStorage.clear();
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'div[data-testid="inline-message"], div[id^="ad-"], div[id^="leaderboard-"], div.expanded-dock, div.pz-ad-box, div[id="top-wrapper"], div[id="bottom-wrapper"]\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'architecturaldigest.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'.paywall-bar, div[class^="MessageBannerWrapper-"\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'bonappetit.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'.paywall-bar, div[class^="MessageBannerWrapper-"\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'cntraveler.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'.paywall-bar, div[class^="MessageBannerWrapper-"\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'epicurious.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'.paywall-bar, div[class^="MessageBannerWrapper-"\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'gq.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'.paywall-bar, div[class^="MessageBannerWrapper-"\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'vanityfair.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'.paywall-bar, div[class^="MessageBannerWrapper-"\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'vogue.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'.paywall-bar, div[class^="MessageBannerWrapper-"\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'americanbanker.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const inlineGate = document.querySelector(\'.inline-gate\');
+                if (inlineGate) {
+                    inlineGate.classList.remove(\'inline-gate\');
+                    const inlineGated = document.querySelectorAll(\'.inline-gated\');
+                    for (const elem of inlineGated) { elem.classList.remove(\'inline-gated\'); }
+                }
+            });
+        '
+    ],
+    'washingtonpost.com' => [
+        'classElementRemove' => ['paywall-overlay'],
+        'fetchStrategies' => 'fetchFromSelenium',
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                let paywall = document.querySelectorAll(\'div[data-qa$="-ad"], div[id="leaderboard-wrapper"], div[data-qa="subscribe-promo"]\');
+                paywall.forEach(el => { el.remove(); });
+                const images = document.querySelectorAll(\'img\');
+                images.forEach(image => { image.parentElement.style.filter = \'\'; });
+                const headimage = document.querySelectorAll(\'div .aspect-custom\');
+                headimage.forEach(image => { image.style.filter = \'\'; });
+            });
+        ',
+        'idElementRemove' => ['wall-bottom-drawer-container']
+    ],
+    'usatoday.com' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const banners = document.querySelectorAll(\'div.roadblock-container, .gnt_nb, [aria-label="advertisement"], div[id="main-frame-error"]\');
+                banners.forEach(el => { el.remove(); });
+            });
+        '
+    ],
+    'stcatharinesstandard.ca' => [
+        'proxy' => true,
+        'idElementRemove' => 'access-offers-modal',
+        'classElementRemove' => 'modal-backdrop',
+        'classAttrRemove' => ' modal-open'
+    ],
+    'medium.com' => [
+        'headers' => [
+            'Referer' => 'https://t.co/x?amp=1',
+            'X-Forwarded-For' => 'none',
+            'User-Agent' => 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36',
+            'Content-Security-Policy' => 'script-src \'self\';'
+        ]
+    ],
+    'tagesspiegel.de' => [
+        'headers' => [
+            'Content-Security-Policy' => 'script-src \'self\';',
+            'User-Agent' => 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36'
+        ],
+        'urlMods' => [
+            'query' => [
+                [
+                    'key' => 'amp',
+                    'value' => '1'
+                ]
+            ]
+        ]
+    ],
+    'nzz.ch' => [
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const paywall = document.querySelector(\'.dynamic-regwall\');
+                if (paywall) {
+                    paywall.remove();
+                }
+            });
+        '
+    ],
+    'demorgen.be' => [
+        'headers' => [
+            'Cookie' => 'isBot=true; authId=1',
+            'User-Agent' => 'Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Googlebot/2.1; Googlebot-News; +http://www.google.com/bot.html) Chrome/121.0.6140.0 Safari/537.36',
+            'X-Forwarded-For' => 'none',
+            'Referer' => 'https://news.google.com'
+        ],
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                // remove paywall items
+                let paywall = document.querySelectorAll(\'script[src*="advertising-cdn.dpgmedia.cloud"], div[data-temptation-position="ARTICLE_BOTTOM"]\');
+                paywall.forEach(el => { el.remove(); });
+                // remove empty advert
+                const advert = document.querySelector(\'div[data-advert-placeholder-collapses]\');
+                if (advert) {
+                    advert.remove();
+                }
+            });
+        '
+    ],
+    'ft.com' => [
+        'cookies' => [
+            'next-flags' => null,
+            'next:ads' => null
+        ],
+        'fromGoogleBot' => true,
+        'headers' => [
+            'Referer' => 'https://t.co/x?amp=1'
+        ],
+        'customCode' => '
+            document.addEventListener("DOMContentLoaded", () => {
+                const styleTags = document.querySelectorAll(\'link[rel="stylesheet"]\');
+                styleTags.forEach(el => {
+                    const href = el.getAttribute(\'href\');
+                    if (href && href.substring(0, 1) === \'/\') {
+                        const updatedHref = href.substring(1).replace(/(https?:\\/\\/.+?)\\/{2,}/, \'$1/\');
+                        el.setAttribute(\'href\', updatedHref);
+                    }
+                });
+                setTimeout(() => {
+                    const cookie = document.querySelectorAll(\'.o-cookie-message, .js-article-ribbon, .o-ads, .o-banner, .o-message, .article__content-sign-up\');
+                    cookie.forEach(el => { el.remove(); });
+                }, 1000);
+            })
+        '
+    ],
+    // Test domain
     'altendorfme.github.io' => [
         'userAgent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36',
         'headers' => [
@@ -366,6 +731,7 @@ return [
             'Cache-Control' => 'no-cache',
             'Pragma' => 'no-cache'
         ],
+        'proxy' => true,
         'idElementRemove' => ['test-id-1', 'paywall'],
         'classElementRemove' => ['test-class-1'],
         'scriptTagRemove' => ['analytics.js', 'test-script.js', 'paywall.js'],

app/data/global_rules.php

@@ -7,6 +7,7 @@
  * using the 'excludeGlobalRules' configuration in domain_rules.php
  */
 return [
+    'proxy' => false,    
     // Classes to be removed from all pages:
     'classElementRemove' => [
         'subscription',

app/dist/icons/moon.svg

@@ -0,0 +1,3 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" fill="currentColor" class="bi bi-moon-fill" viewBox="0 0 16 16">
+  <path d="M6 .278a.77.77 0 0 1 .08.858 7.2 7.2 0 0 0-.878 3.46c0 4.021 3.278 7.277 7.318 7.277q.792-.001 1.533-.16a.79.79 0 0 1 .81.316.73.73 0 0 1-.031.893A8.35 8.35 0 0 1 8.344 16C3.734 16 0 12.286 0 7.71 0 4.266 2.114 1.312 5.124.06A.75.75 0 0 1 6 .278"/>
+</svg>

app/dist/icons/sun.svg

@@ -0,0 +1,3 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" fill="currentColor" class="bi bi-brightness-high-fill" viewBox="0 0 16 16">
+  <path d="M12 8a4 4 0 1 1-8 0 4 4 0 0 1 8 0M8 0a.5.5 0 0 1 .5.5v2a.5.5 0 0 1-1 0v-2A.5.5 0 0 1 8 0m0 13a.5.5 0 0 1 .5.5v2a.5.5 0 0 1-1 0v-2A.5.5 0 0 1 8 13m8-5a.5.5 0 0 1-.5.5h-2a.5.5 0 0 1 0-1h2a.5.5 0 0 1 .5.5M3 8a.5.5 0 0 1-.5.5h-2a.5.5 0 0 1 0-1h2A.5.5 0 0 1 3 8m10.657-5.657a.5.5 0 0 1 0 .707l-1.414 1.415a.5.5 0 1 1-.707-.708l1.414-1.414a.5.5 0 0 1 .707 0m-9.193 9.193a.5.5 0 0 1 0 .707L3.05 13.657a.5.5 0 0 1-.707-.707l1.414-1.414a.5.5 0 0 1 .707 0m9.193 2.121a.5.5 0 0 1-.707 0l-1.414-1.414a.5.5 0 0 1 .707-.707l1.414 1.414a.5.5 0 0 1 0 .707M4.464 4.465a.5.5 0 0 1-.707 0L2.343 3.05a.5.5 0 1 1 .707-.707l1.414 1.414a.5.5 0 0 1 0 .708"/>
+</svg>

app/dist/js/scripts.js

@@ -1,2 +1,2 @@
-"serviceWorker"in navigator&&window.addEventListener("load",()=>{navigator.serviceWorker.register("/service-worker.js").then(()=>{}).catch(()=>{})}),document.addEventListener("DOMContentLoaded",function(){let t=document.querySelector(".integration");var e=document.querySelector(".integration__toggle");let o=document.querySelector(".extension");var n=document.querySelector(".extension__toggle");let r=e=>{e!==t&&t.classList.remove("open"),e!==o&&o.classList.remove("open")};e.addEventListener("click",e=>{e.stopPropagation(),r(t),t.classList.toggle("open")}),n.addEventListener("click",e=>{e.stopPropagation(),r(o),o.classList.toggle("open")}),t.addEventListener("click",e=>{e.stopPropagation()}),o.addEventListener("click",e=>{e.stopPropagation()}),document.addEventListener("click",()=>{t.classList.remove("open"),o.classList.remove("open")}),document.addEventListener("click",e=>{e=e.target.closest(".toasty");e&&e.remove()}),document.addEventListener("click",e=>{e.target.closest(".open-nav")&&((e=document.querySelector("header")).classList.contains("open")?e.classList.remove("open"):e.classList.add("open"))});e=document.getElementById("paste");let a=document.getElementById("url");e&&a&&e.addEventListener("click",async e=>{e.preventDefault();try{var t=await navigator.clipboard.readText();a.value=t.trim()}catch(e){console.error("Failed to read clipboard contents",e)}})});
+"serviceWorker"in navigator&&window.addEventListener("load",()=>{navigator.serviceWorker.register("/service-worker.js").then(()=>{}).catch(()=>{})}),document.addEventListener("DOMContentLoaded",function(){let t=document.querySelector(".integration");var e=document.querySelector(".integration__toggle");let o=document.querySelector(".extension");var n=document.querySelector(".extension__toggle");let a=e=>{e!==t&&t.classList.remove("open"),e!==o&&o.classList.remove("open")};e.addEventListener("click",e=>{e.stopPropagation(),a(t),t.classList.toggle("open")}),n.addEventListener("click",e=>{e.stopPropagation(),a(o),o.classList.toggle("open")}),t.addEventListener("click",e=>{e.stopPropagation()}),o.addEventListener("click",e=>{e.stopPropagation()}),document.addEventListener("click",()=>{t.classList.remove("open"),o.classList.remove("open")}),document.addEventListener("click",e=>{e=e.target.closest(".toasty");e&&e.remove()}),document.addEventListener("click",e=>{e.target.closest(".open-nav")&&((e=document.querySelector("header")).classList.contains("open")?e.classList.remove("open"):e.classList.add("open"))});e=document.getElementById("paste");let r=document.getElementById("url");e&&r&&e.addEventListener("click",async e=>{e.preventDefault();try{var t=await navigator.clipboard.readText();r.value=t.trim()}catch(e){console.error("Failed to read clipboard contents",e)}});n=document.getElementById("themeToggle");let c=document.documentElement;e=localStorage.getItem("theme")||"light";c.setAttribute("data-theme",e),n&&n.addEventListener("click",()=>{var e="dark"===c.getAttribute("data-theme")?"light":"dark";c.setAttribute("data-theme",e),localStorage.setItem("theme",e)})});
 //# sourceMappingURL=scripts.js.map

app/inc/Rules.php

@@ -31,7 +31,9 @@ class Rules
         'fetchStrategies',
         'fromGoogleBot',
         'removeElementsByTag',
-        'removeCustomAttr'
+        'removeCustomAttr',
+        'urlMods',
+        'proxy'
     ];
 
     /**
@@ -91,6 +93,35 @@ class Rules
         return $this->getGlobalRules();
     }
 
+
+    /**
+     * Retrieves merged rules for a domain
+     * @param string $domain Target domain
+     * @return array|null Combined ruleset or global rules
+     */
+    public function hasDomainRules($domain)
+    {
+        $domainParts = $this->getDomainParts($domain);
+
+        // Check for exact domain match first
+        foreach ($this->domainRules as $pattern => $rules) {
+            if ($this->getBaseDomain($domain) === $this->getBaseDomain($pattern)) {
+                return true;
+            }
+        }
+
+        // Check for partial domain matches
+        foreach ($domainParts as $part) {
+            foreach ($this->domainRules as $pattern => $rules) {
+                if ($part === $this->getBaseDomain($pattern)) {
+                    return true;
+                }
+            }
+        }
+
+        return false;
+    }
+
     /**
      * Combines domain rules with global configuration
      * @param array $rules Domain-specific rules
@@ -108,12 +139,14 @@ class Rules
 
             if (isset($excludeGlobalRules[$ruleType])) {
                 if (is_assoc_array($globalTypeRules)) {
-                    $mergedRules[$ruleType] = array_diff_key($globalTypeRules, array_flip($excludeGlobalRules[$ruleType]));
+                    $result = array_diff_key($globalTypeRules, array_flip($excludeGlobalRules[$ruleType]));
+                    $mergedRules[$ruleType] = is_array($result) ? $result : [];
                 } else {
-                    $mergedRules[$ruleType] = array_diff($globalTypeRules, $excludeGlobalRules[$ruleType]);
+                    $result = array_diff($globalTypeRules, $excludeGlobalRules[$ruleType]);
+                    $mergedRules[$ruleType] = is_array($result) ? $result : [];
                 }
             } else {
-                $mergedRules[$ruleType] = $globalTypeRules;
+                $mergedRules[$ruleType] = is_array($globalTypeRules) ? $globalTypeRules : [];
             }
         }
 
@@ -126,10 +159,13 @@ class Rules
             }
 
             if (in_array($ruleType, ['cookies', 'headers'])) {
-                $mergedRules[$ruleType] = array_merge($mergedRules[$ruleType], $domainTypeRules);
+                $mergedRules[$ruleType] = array_merge(
+                    is_array($mergedRules[$ruleType]) ? $mergedRules[$ruleType] : [],
+                    is_array($domainTypeRules) ? $domainTypeRules : []
+                );
             } else {
                 $mergedRules[$ruleType] = array_values(array_unique(array_merge(
-                    $mergedRules[$ruleType],
+                    is_array($mergedRules[$ruleType]) ? $mergedRules[$ruleType] : [],
                     (array)$domainTypeRules
                 )));
             }

app/inc/URLAnalyzer.php

@@ -50,6 +50,31 @@ class URLAnalyzer extends URLAnalyzerBase
      */
     public function analyze($url)
     {
+        // Extract and validate hostname
+        $host = parse_url($url, PHP_URL_HOST);
+        if (!$host) {
+            $this->error->throwError(self::ERROR_INVALID_URL, '');
+        }
+
+        // Check if URL contains restricted keywords
+        if ($this->isRestrictedUrl($url)) {
+            Logger::getInstance()->logUrl($url, 'RESTRICTED_URL');
+            $this->error->throwError(self::ERROR_RESTRICTED_URL, '');
+        }
+        $originalHost = parse_url($url, PHP_URL_HOST);
+        $host = preg_replace('/^www\./', '', $host);
+
+        // Check if domain is in DMCA list FIRST (before any HTTP requests)
+        foreach (DMCA_DOMAINS as $dmcaEntry) {
+            if (is_array($dmcaEntry) && isset($dmcaEntry['host'])) {
+                if ($dmcaEntry['host'] === $host || $dmcaEntry['host'] === $originalHost) {
+                    Logger::getInstance()->logUrl($url, 'DMCA_DOMAIN');
+                    $customMessage = isset($dmcaEntry['message']) ? $dmcaEntry['message'] : '';
+                    $this->error->throwError(self::ERROR_DMCA_DOMAIN, $customMessage);
+                }
+            }
+        }
+
         // Reset activated rules for new analysis
         $this->activatedRules = [];
 
@@ -60,20 +85,17 @@ class URLAnalyzer extends URLAnalyzerBase
             return $this->process->processContent($rawContent, parse_url($url, PHP_URL_HOST), $url);
         }
 
-        // Extract and validate hostname
-        $host = parse_url($url, PHP_URL_HOST);
-        if (!$host) {
-            $this->error->throwError(self::ERROR_INVALID_URL, '');
-        }
-        $host = preg_replace('/^www\./', '', $host);
-
         // Check if domain is in blocked list
         if (in_array($host, BLOCKED_DOMAINS)) {
             Logger::getInstance()->logUrl($url, 'BLOCKED_DOMAIN');
             $this->error->throwError(self::ERROR_BLOCKED_DOMAIN, '');
         }
 
-        // Check HTTP status and handle any errors
+        // Check if domain has specific rules by looking for domain-specific configurations
+        $hasCustomRules = $this->hasDomainRules($host);
+        
+        // Check HTTP status and handle any errors only if domain doesn't have custom rules
+        if (!$hasCustomRules) {
             $redirectInfo = $this->utils->checkStatus($url);
             if ($redirectInfo['httpCode'] !== 200) {
                 Logger::getInstance()->logUrl($url, 'INVALID_STATUS_CODE', "HTTP {$redirectInfo['httpCode']}");
@@ -83,6 +105,7 @@ class URLAnalyzer extends URLAnalyzerBase
                     $this->error->throwError(self::ERROR_HTTP_ERROR, (string)$redirectInfo['httpCode']);
                 }
             }
+        }
 
         try {
             // Get specific rules for this domain

app/inc/URLAnalyzer/URLAnalyzerBase.php

@@ -22,23 +22,27 @@ class URLAnalyzerBase
     /** @var string Error constants for different failure scenarios */
     const ERROR_INVALID_URL = 'INVALID_URL';
     const ERROR_BLOCKED_DOMAIN = 'BLOCKED_DOMAIN';
+    const ERROR_DMCA_DOMAIN = 'DMCA_DOMAIN';
     const ERROR_NOT_FOUND = 'NOT_FOUND';
     const ERROR_HTTP_ERROR = 'HTTP_ERROR';
     const ERROR_CONNECTION_ERROR = 'CONNECTION_ERROR';
     const ERROR_DNS_FAILURE = 'DNS_FAILURE';
     const ERROR_CONTENT_ERROR = 'CONTENT_ERROR';
     const ERROR_GENERIC_ERROR = 'GENERIC_ERROR';
+    const ERROR_RESTRICTED_URL = 'RESTRICTED_URL';
 
     /** @var array Maps error types to HTTP codes and message keys */
     protected $errorMap = [
         self::ERROR_INVALID_URL => ['code' => 400, 'message_key' => 'INVALID_URL'],
         self::ERROR_BLOCKED_DOMAIN => ['code' => 403, 'message_key' => 'BLOCKED_DOMAIN'],
+        self::ERROR_DMCA_DOMAIN => ['code' => 403, 'message_key' => 'DMCA_DOMAIN'],
         self::ERROR_NOT_FOUND => ['code' => 404, 'message_key' => 'NOT_FOUND'],
         self::ERROR_HTTP_ERROR => ['code' => 502, 'message_key' => 'HTTP_ERROR'],
         self::ERROR_CONNECTION_ERROR => ['code' => 503, 'message_key' => 'CONNECTION_ERROR'],
         self::ERROR_DNS_FAILURE => ['code' => 504, 'message_key' => 'DNS_FAILURE'],
         self::ERROR_CONTENT_ERROR => ['code' => 502, 'message_key' => 'CONTENT_ERROR'],
-        self::ERROR_GENERIC_ERROR => ['code' => 500, 'message_key' => 'GENERIC_ERROR']
+        self::ERROR_GENERIC_ERROR => ['code' => 500, 'message_key' => 'GENERIC_ERROR'],
+        self::ERROR_RESTRICTED_URL => ['code' => 403, 'message_key' => 'RESTRICTED_URL']
     ];
 
     /** @var array List of user agents to rotate through, including Googlebot */
@@ -113,4 +117,59 @@ class URLAnalyzerBase
     {
         return $this->rules->getDomainRules($domain);
     }
+    
+    /**
+     * Check if domain has specific rules
+     * @param string $host The domain host to check
+     * @return bool True if domain has custom rules, false otherwise
+     */
+    protected function hasDomainRules($domain)
+    {
+        return $this->rules->hasDomainRules($domain);
+    }
+
+    /**
+     * Check if URL contains restricted keywords
+     * @param string $url The URL to check
+     * @return bool True if URL contains restricted keywords, false otherwise
+     */
+    protected function isRestrictedUrl($url)
+    {
+        $restrictedKeywords = [
+            'login',
+            'signin',
+            'sign-in',
+            'signup',
+            'sign-up',
+            'register',
+            'registration',
+            'lost-password',
+            'forgot-password',
+            'reset-password',
+            'password',
+            'auth',
+            'authentication',
+            'account',
+            'profile',
+            'dashboard',
+            'admin',
+            'member',
+            'subscription',
+            'subscribe',
+            'premium',
+            'checkout',
+            'payment',
+            'billing'
+        ];
+
+        $urlLower = strtolower($url);
+        
+        foreach ($restrictedKeywords as $keyword) {
+            if (strpos($urlLower, $keyword) !== false) {
+                return true;
+            }
+        }
+        
+        return false;
+    }
 }

app/inc/URLAnalyzer/URLAnalyzerError.php

@@ -14,10 +14,17 @@ class URLAnalyzerError extends URLAnalyzerBase
     public function throwError($errorType, $additionalInfo = '')
     {
         $errorConfig = $this->errorMap[$errorType];
+        
+        // For DMCA domains, use custom message if provided, otherwise use default
+        if ($errorType === self::ERROR_DMCA_DOMAIN && !empty($additionalInfo)) {
+            $message = $additionalInfo;
+        } else {
             $message = Language::getMessage($errorConfig['message_key'])['message'];
-        if ($additionalInfo) {
+            if ($additionalInfo && $errorType !== self::ERROR_DMCA_DOMAIN) {
                 $message .= ': ' . $additionalInfo;
             }
+        }
+        
         throw new URLAnalyzerException($message, $errorConfig['code'], $errorType, $additionalInfo);
     }
 }

app/inc/URLAnalyzer/URLAnalyzerFetch.php

@@ -21,16 +21,117 @@ class URLAnalyzerFetch extends URLAnalyzerBase
     /**
      * Sets up the fetch handler with error handling capability
      */
+    /** @var array List of available proxies */
+    private $proxyList = [];
+    
+    /** @var string Path to proxy cache file */
+    private $proxyCachePath = '';
+    
     public function __construct()
     {
         parent::__construct();
         $this->error = new URLAnalyzerError();
+        $this->proxyCachePath = __DIR__ . '/../../cache/proxy_list.json';
+        $this->loadProxyList();
+    }
+    
+    /**
+     * Loads proxy list from cache if available
+     */
+    private function loadProxyList()
+    {
+        if (isset($_ENV['PROXY_LIST']) && file_exists($this->proxyCachePath)) {
+            $cachedList = file_get_contents($this->proxyCachePath);
+            if (!empty($cachedList)) {
+                $this->proxyList = json_decode($cachedList, true);
+            }
+        }
+    }
+    
+    /**
+     * Gets a random proxy from the list
+     * @return string|null Random proxy URL or null if none available
+     */
+    private function getRandomProxy()
+    {
+        if (empty($this->proxyList)) {
+            return null;
+        }
+        
+        return $this->proxyList[array_rand($this->proxyList)];
     }
 
     /** 
      * Fetches content using cURL
      * Handles redirects and custom headers
      */
+    /**
+     * Modifies URL based on urlMods rules
+     * @param string $url Original URL
+     * @param array $domainRules Domain rules containing urlMods
+     * @return string Modified URL
+     */
+    private function applyUrlModifications($url, $domainRules)
+    {
+        if (!isset($domainRules['urlMods'])) {
+            return $url;
+        }
+
+        $urlParts = parse_url($url);
+        
+        if (isset($domainRules['urlMods']['query']) && is_array($domainRules['urlMods']['query'])) {
+            $queryParams = [];
+            
+            if (isset($urlParts['query'])) {
+                parse_str($urlParts['query'], $queryParams);
+            }
+            
+            foreach ($domainRules['urlMods']['query'] as $queryMod) {
+                if (isset($queryMod['key']) && isset($queryMod['value'])) {
+                    $queryParams[$queryMod['key']] = $queryMod['value'];
+                }
+            }
+            
+            $urlParts['query'] = http_build_query($queryParams);
+        }
+
+        $modifiedUrl = '';
+        
+        if (isset($urlParts['scheme'])) {
+            $modifiedUrl .= $urlParts['scheme'] . '://';
+        }
+        
+        if (isset($urlParts['user'])) {
+            $modifiedUrl .= $urlParts['user'];
+            if (isset($urlParts['pass'])) {
+                $modifiedUrl .= ':' . $urlParts['pass'];
+            }
+            $modifiedUrl .= '@';
+        }
+        
+        if (isset($urlParts['host'])) {
+            $modifiedUrl .= $urlParts['host'];
+        }
+        
+        if (isset($urlParts['port'])) {
+            $modifiedUrl .= ':' . $urlParts['port'];
+        }
+        
+        if (isset($urlParts['path'])) {
+            $modifiedUrl .= $urlParts['path'];
+        }
+        
+        if (isset($urlParts['query'])) {
+            $modifiedUrl .= '?' . $urlParts['query'];
+        }
+        
+        if (isset($urlParts['fragment'])) {
+            $modifiedUrl .= '#' . $urlParts['fragment'];
+        }
+        
+        return $modifiedUrl;
+    }
+
     public function fetchContent($url)
     {
         $curl = new Curl();
@@ -42,6 +143,8 @@ class URLAnalyzerFetch extends URLAnalyzerBase
         $host = preg_replace('/^www\./', '', $host);
         $domainRules = $this->getDomainRules($host);
         
+        $url = $this->applyUrlModifications($url, $domainRules);
+
         $curl->setOpt(CURLOPT_FOLLOWLOCATION, true);
         $curl->setOpt(CURLOPT_MAXREDIRS, 2);
         $curl->setOpt(CURLOPT_TIMEOUT, 10);
@@ -49,6 +152,13 @@ class URLAnalyzerFetch extends URLAnalyzerBase
         $curl->setOpt(CURLOPT_DNS_SERVERS, implode(',', $this->dnsServers));
         $curl->setOpt(CURLOPT_ENCODING, '');
         
+        if (isset($domainRules['proxy']) && $domainRules['proxy'] === true) {
+            $proxy = $this->getRandomProxy();
+            if ($proxy) {
+                $curl->setOpt(CURLOPT_PROXY, $proxy);
+            }
+        }
+
         $curl->setHeaders([
             'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',
             'Accept-Language' => 'en-US,en;q=0.5',
@@ -97,6 +207,13 @@ class URLAnalyzerFetch extends URLAnalyzerBase
      */
     public function fetchFromWaybackMachine($url)
     {
+        $domainHost = parse_url($url, PHP_URL_HOST);
+        if ($domainHost) {
+            $domainHost = preg_replace('/^www\./', '', $domainHost);
+            $domainRules = $this->getDomainRules($domainHost);
+            $url = $this->applyUrlModifications($url, $domainRules);
+        }
+        
         $url = preg_replace('#^https?://#', '', $url);
         $availabilityUrl = "https://archive.org/wayback/available?url=" . urlencode($url);
 
@@ -106,6 +223,13 @@ class URLAnalyzerFetch extends URLAnalyzerBase
         $curl->setOpt(CURLOPT_SSL_VERIFYPEER, false);
         $curl->setUserAgent($this->getRandomUserAgent());
         
+        if (isset($domainRules['proxy']) && $domainRules['proxy'] === true) {
+            $proxy = $this->getRandomProxy();
+            if ($proxy) {
+                $curl->setOpt(CURLOPT_PROXY, $proxy);
+            }
+        }
+
         $curl->get($availabilityUrl);
 
         if ($curl->error) {
@@ -130,6 +254,13 @@ class URLAnalyzerFetch extends URLAnalyzerBase
         $curl->setOpt(CURLOPT_SSL_VERIFYPEER, false);
         $curl->setUserAgent($this->getRandomUserAgent());
         
+        if (isset($domainRules['proxy']) && $domainRules['proxy'] === true) {
+            $proxy = $this->getRandomProxy();
+            if ($proxy) {
+                $curl->setOpt(CURLOPT_PROXY, $proxy);
+            }
+        }
+
         $curl->get($archiveUrl);
 
         if ($curl->error || $curl->httpStatusCode !== 200 || empty($curl->response)) {
@@ -152,16 +283,32 @@ class URLAnalyzerFetch extends URLAnalyzerBase
     {
         $host = 'http://'.SELENIUM_HOST.'/wd/hub';
         
+        $domainHost = parse_url($url, PHP_URL_HOST);
+        if ($domainHost) {
+            $domainHost = preg_replace('/^www\./', '', $domainHost);
+            $domainRules = $this->getDomainRules($domainHost);
+            $url = $this->applyUrlModifications($url, $domainRules);
+        }
+
+        $useProxy = isset($domainRules['proxy']) && $domainRules['proxy'] === true;
+        $proxy = $useProxy ? $this->getRandomProxy() : null;
+
         if ($browser === 'chrome') {
             $options = new ChromeOptions();
-            $options->addArguments([
+            $arguments = [
                 '--headless',
                 '--disable-gpu',
                 '--no-sandbox',
                 '--disable-dev-shm-usage',
                 '--disable-images',
                 '--blink-settings=imagesEnabled=false'
-            ]);
+            ];
+            
+            if ($useProxy && $proxy) {
+                $arguments[] = '--proxy-server=' . $proxy;
+            }
+            
+            $options->addArguments($arguments);
 
             $capabilities = DesiredCapabilities::chrome();
             $capabilities->setCapability(ChromeOptions::CAPABILITY, $options);
@@ -174,6 +321,22 @@ class URLAnalyzerFetch extends URLAnalyzerBase
             $profile->setPreference("network.http.referer.spoofSource", true);
             $profile->setPreference("network.http.referer.trimmingPolicy", 0);
             
+            if ($useProxy && $proxy) {
+                $proxyParts = parse_url($proxy);
+                if (isset($proxyParts['host']) && isset($proxyParts['port'])) {
+                    $profile->setPreference("network.proxy.type", 1);
+                    $profile->setPreference("network.proxy.http", $proxyParts['host']);
+                    $profile->setPreference("network.proxy.http_port", $proxyParts['port']);
+                    $profile->setPreference("network.proxy.ssl", $proxyParts['host']);
+                    $profile->setPreference("network.proxy.ssl_port", $proxyParts['port']);
+                    
+                    if (isset($proxyParts['user']) && isset($proxyParts['pass'])) {
+                        $profile->setPreference("network.proxy.username", $proxyParts['user']);
+                        $profile->setPreference("network.proxy.password", $proxyParts['pass']);
+                    }
+                }
+            }
+
             $options = new FirefoxOptions();
             $options->setProfile($profile);
 

app/inc/URLAnalyzer/URLAnalyzerUtils.php

@@ -18,7 +18,17 @@ class URLAnalyzerUtils extends URLAnalyzerBase
         $curl->setOpt(CURLOPT_TIMEOUT, 5);
         $curl->setOpt(CURLOPT_SSL_VERIFYPEER, false);
         $curl->setOpt(CURLOPT_NOBODY, true);
-        $curl->setUserAgent($this->getRandomUserAgent());
+        $curl->setOpt(CURLOPT_DNS_SERVERS, '8.8.8.8,8.4.4.8');
+        $curl->setHeaders([
+            'User-Agent' => 'Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/W.X.Y.Z Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)',
+            'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',
+            'Accept-Language' => 'en-US,en;q=0.5',
+            'Cache-Control' => 'no-cache',
+            'Pragma' => 'no-cache',
+            'DNT' => '1',
+            'X-Forwarded-For' => '66.249.' . rand(64, 95) . '.' . rand(1, 254),
+            'From' => 'googlebot(at)googlebot.com'
+        ]);
         $curl->get($url);
 
         if ($curl->error) {

app/languages/de-de.php

@@ -21,6 +21,10 @@ return [
             'message' => 'Diese Seite ist nicht erlaubt.',
             'type' => 'error'
         ],
+        'DMCA_DOMAIN' => [
+            'message' => 'Die angeforderte Website kann aufgrund von Anfragen ihrer Eigentümer nicht angezeigt werden.',
+            'type' => 'error'
+        ],
         'DNS_FAILURE' => [
             'message' => 'DNS für die Domain konnte nicht aufgelöst werden. Bitte überprüfe, ob die URL korrekt ist.',
             'type' => 'warning'
@@ -48,6 +52,10 @@ return [
         'GENERIC_ERROR' => [
             'message' => 'Bei der Bearbeitung Ihrer Anfrage ist ein Fehler aufgetreten.',
             'type' => 'warning'
+        ],
+        'RESTRICTED_URL' => [
+            'message' => 'Diese URL enthält eingeschränkten Inhalt und kann aus Sicherheitsgründen nicht verarbeitet werden.',
+            'type' => 'error'
         ]
     ]
 ];

app/languages/en.php

@@ -21,6 +21,10 @@ return [
             'message' => 'This domain is blocked for extraction.',
             'type' => 'error'
         ],
+        'DMCA_DOMAIN' => [
+            'message' => 'The requested website cannot be displayed due to requests from its owners.',
+            'type' => 'error'
+        ],
         'DNS_FAILURE' => [
             'message' => 'Failed to resolve DNS for the domain. Please verify if the URL is correct.',
             'type' => 'warning'
@@ -48,6 +52,10 @@ return [
         'GENERIC_ERROR' => [
             'message' => 'An error occurred while processing your request.',
             'type' => 'warning'
+        ],
+        'RESTRICTED_URL' => [
+            'message' => 'This URL contains restricted content and cannot be processed for security reasons.',
+            'type' => 'error'
         ]
     ]
 ];

app/languages/es.php

@@ -21,6 +21,10 @@ return [
             'message' => 'Este dominio está bloqueado para extracción.',
             'type' => 'error'
         ],
+        'DMCA_DOMAIN' => [
+            'message' => 'El sitio web solicitado no se puede mostrar debido a las solicitudes de sus propietarios.',
+            'type' => 'error'
+        ],
         'DNS_FAILURE' => [
             'message' => 'Error al resolver DNS para el dominio. Verifique si la URL es correcta.',
             'type' => 'warning'
@@ -48,6 +52,10 @@ return [
         'GENERIC_ERROR' => [
             'message' => 'Ocurrió un error al procesar su solicitud.',
             'type' => 'warning'
+        ],
+        'RESTRICTED_URL' => [
+            'message' => 'Esta URL contiene contenido restringido y no se puede procesar por razones de seguridad.',
+            'type' => 'error'
         ]
     ]
 ];

app/languages/pt-br.php

@@ -21,6 +21,10 @@ return [
             'message' => 'Este domínio está bloqueado para extração.',
             'type' => 'error'
         ],
+        'DMCA_DOMAIN' => [
+            'message' => 'O site solicitado não pode ser exibido por exigência dos seus proprietários.',
+            'type' => 'error'
+        ],
         'DNS_FAILURE' => [
             'message' => 'Falha ao resolver DNS para o domínio. Verifique se a URL está correta.',
             'type' => 'warning'
@@ -48,6 +52,10 @@ return [
         'GENERIC_ERROR' => [
             'message' => 'Ocorreu um erro ao processar sua solicitação.',
             'type' => 'warning'
+        ],
+        'RESTRICTED_URL' => [
+            'message' => 'Esta URL contém conteúdo restrito e não pode ser processada por motivos de segurança.',
+            'type' => 'error'
         ]
     ]
 ];

app/languages/ru-ru.php

@@ -21,6 +21,10 @@ return [
             'message' => 'Этот домен заблокирован для извлечения.',
             'type' => 'error'
         ],
+        'DMCA_DOMAIN' => [
+            'message' => 'Запрошенный веб-сайт не может быть отображен по запросу его владельцев.',
+            'type' => 'error'
+        ],
         'DNS_FAILURE' => [
             'message' => 'Не удалось разрешить DNS для домена. Проверьте правильность URL.',
             'type' => 'warning'
@@ -48,6 +52,10 @@ return [
         'GENERIC_ERROR' => [
             'message' => 'При обработке вашего запроса произошла ошибка.',
             'type' => 'warning'
+        ],
+        'RESTRICTED_URL' => [
+            'message' => 'Этот URL-адрес содержит запрещенный контент и не может быть обработан по соображениям безопасности.',
+            'type' => 'error'
         ]
     ]
 ];

app/src/URLProcessor.php

@@ -5,6 +5,7 @@ namespace App;
 use Inc\Language;
 use Inc\URLAnalyzer;
 use Inc\URLAnalyzer\URLAnalyzerException;
+use Inc\Cache;
 
 /**
  * URL Processor
@@ -109,6 +110,19 @@ class URLProcessor
             } else {
                 if ($errorType === URLAnalyzer::ERROR_BLOCKED_DOMAIN && $additionalInfo) {
                     $this->redirect(trim($additionalInfo), $errorType);
+                } elseif ($errorType === URLAnalyzer::ERROR_DMCA_DOMAIN) {
+                    // For DMCA domains, show the custom message directly instead of redirecting
+                    Language::init(LANGUAGE);
+                    $message = $e->getMessage();
+                    $message_type = 'error';
+                    $url = ''; // Initialize url variable for the view
+                    
+                    // Initialize cache for counting
+                    $cache = new \Inc\Cache();
+                    $cache_folder = $cache->getCacheFileCount();
+                    
+                    require __DIR__ . '/views/home.php';
+                    exit;
                 }
                 $this->redirect(SITE_URL, $errorType);
             }

app/src/views/home.php

@@ -60,6 +60,7 @@
                     </div>
                 </div>
             </nav>
+            <div class="fast_buttons">
                 <div class="extension">
                     <button class="extension__toggle"><?php echo \Inc\Language::get('nav_extension'); ?></button>
                     <div class="extension__menu">
@@ -67,6 +68,13 @@
                         <a target="_blank" href="https://chromewebstore.google.com/detail/marreta/ipelapagohjgjcgpncpbmaaacemafppe"><span class="name">Chrome</span><span class="icon icon--chrome"></span></a>
                     </div>
                 </div>
+                <div class="theme-controls">
+                    <button class="theme-toggle" id="themeToggle">
+                        <span class="icon icon--sun"></span>
+                        <span class="icon icon--moon"></span>
+                    </button>
+                </div>
+            </div>
         </header>
 
         <main>
@@ -85,7 +93,8 @@
                             value="<?php echo htmlspecialchars($url); ?>"
                             required
                             pattern="https?://.+"
-                            title="<?php echo \Inc\Language::getMessage('INVALID_URL')['message']; ?>">
+                            title="<?php echo \Inc\Language::getMessage('INVALID_URL')['message']; ?>"
+                            autofocus>
                             <span class="paste" id="paste"><span class="icon icon--paste"></span></span>
                     </div>
                     <button type="submit" alt="<?php echo \Inc\Language::get('analyze_button'); ?>">

docker-compose.yml

@@ -16,6 +16,8 @@ services:
       - LANGUAGE=${LANGUAGE:-pt-br}
       - LOG_LEVEL=${LOG_LEVEL:-WARNING}
       - SELENIUM_HOST=${SELENIUM_HOST:-selenium-hub:4444}
+      - CLEANUP_DAYS=7 # Optional
+      - PROXY_LIST=url # Optional
     restart: unless-stopped
   # Selenium
   selenium-hub:

docker-entrypoint.sh

@@ -109,8 +109,21 @@ nginx -g "daemon off;" &
 sleep 3
 check_nginx
 
+# Starting Cron
+log_info "Starting Cron..."
+service cron restart
+log_success "Cron started"
+
 echo -e "\n${GREEN}=== Marreta initialized ===${NC}\n"
 
+# Run proxy list updater
+log_info "Running proxy list updater..."
+if php /app/bin/proxy; then
+    log_success "Proxy list updater completed successfully"
+else
+    log_info "Proxy list updater finished (may not have been configured)"
+fi
+
 # Wait for any process to exit
 wait -n