From 01b903da88297db725b5a195a615d34f707c4901 Mon Sep 17 00:00:00 2001
From: Renan Bernordi <altendorfme@gmail.com>
Date: Thu, 5 Dec 2024 19:08:16 -0300
Subject: [PATCH] novos sites bloquedos e melhoria no validador

---
 app/data/blocked_domains.php | 10 +++++++++-
 app/inc/URLAnalyzer.php      |  4 ++--
 2 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/app/data/blocked_domains.php b/app/data/blocked_domains.php
index 33629b9..eb77cf3 100644
--- a/app/data/blocked_domains.php
+++ b/app/data/blocked_domains.php
@@ -103,5 +103,13 @@ return [
     'archive.today',
     '12ft.io',
     'jusbrasil.com.br',
-    'glassdoor.com.br'
+    'glassdoor.com.br',
+    'gov.br',
+    'medium.com',
+    'stackoverflow.com',
+    'hoteis.com',
+    'amazon.com',
+    'amazon.com.br',
+    'msn.com',
+    'archive.ph'
 ];
diff --git a/app/inc/URLAnalyzer.php b/app/inc/URLAnalyzer.php
index 2f8f1b8..f74013c 100644
--- a/app/inc/URLAnalyzer.php
+++ b/app/inc/URLAnalyzer.php
@@ -110,8 +110,8 @@ class URLAnalyzer
             return $this->cache->get($cleanUrl);
         }
 
-        $parsedUrl = parse_url($cleanUrl);
-        $domain = $parsedUrl['host'];
+        $domain = parse_url($cleanUrl, PHP_URL_HOST);
+        $domain = preg_replace('/^www\./', '', $domain);
 
         // Verificação de domínios bloqueados
         foreach (BLOCKED_DOMAINS as $blockedDomain) {