From c9e59dfce20d10778d1455fb80f47d886a320cf3 Mon Sep 17 00:00:00 2001
From: Renan Bernordi <altendorfme@gmail.com>
Date: Tue, 24 Dec 2024 13:35:20 -0300
Subject: [PATCH] removido logs e liberado selenium como ultimo fallback

---
 TESTED_URLS.md               |  2 ++
 app/data/blocked_domains.php |  1 +
 app/inc/URLAnalyzer.php      | 17 ++++++++++++++---
 3 files changed, 17 insertions(+), 3 deletions(-)

diff --git a/TESTED_URLS.md b/TESTED_URLS.md
index 27b7096..543db23 100644
--- a/TESTED_URLS.md
+++ b/TESTED_URLS.md
@@ -39,6 +39,8 @@ https://oantagonista.com.br/brasil/lewandowski-insiste-na-pec-da-seguranca/
 https://jornaldebrasilia.com.br/noticias/politica-e-poder/lula-aguarda-pt-para-troca-em-pastas-chefiadas-por-petistas-em-reforma-ministerial/
 https://opopular.com.br/cidades/ex-secretario-de-saude-de-goiania-deixa-hospital-e-volta-para-a-cadeia-1.3207162
 https://www.cartacapital.com.br/politica/surpresa-natalina/
+https://seucreditodigital.com.br/123milhas-devera-apresentar-plano-de-recuperacao-ainda-este-mes/
+https://www.matinaljornalismo.com.br/matinal/reportagem-matinal/vazao-guaiba-porto-alegre/
 
 ## Internacional
 https://www.nytimes.com/2024/11/20/us/politics/matt-gaetz-venmo-payments-sex.html
diff --git a/app/data/blocked_domains.php b/app/data/blocked_domains.php
index 73ece58..3d0fcb4 100644
--- a/app/data/blocked_domains.php
+++ b/app/data/blocked_domains.php
@@ -16,6 +16,7 @@
     'utppublishing.com',
     'chronicle.com',
     'nexojornal.com',
+    'nexojornal.com.br',
     'lesoir.be',
     'weeklytimesnow.com.au',
     'barrons.com',
diff --git a/app/inc/URLAnalyzer.php b/app/inc/URLAnalyzer.php
index 13c37c6..1166667 100644
--- a/app/inc/URLAnalyzer.php
+++ b/app/inc/URLAnalyzer.php
@@ -123,7 +123,6 @@ public function analyze($url)
         $host = preg_replace('/^www\./', '', $host);
 
         if (in_array($host, BLOCKED_DOMAINS)) {
-            Logger::getInstance()->log($cleanUrl, 'BLOCKED_DOMAIN');
             throw new Exception('Este domínio está bloqueado para extração.');
         }
 
@@ -153,7 +152,7 @@ public function analyze($url)
                 return $processedContent;
             }
         } catch (Exception $e) {
-            Logger::getInstance()->log($cleanUrl, 'DIRECT_FETCH_ERROR', $e->getMessage());
+            error_log("DIRECT_FETCH_ERROR: " . $e->getMessage());
         }
 
         // 6. Tenta buscar do Wayback Machine como fallback
@@ -165,7 +164,19 @@ public function analyze($url)
                 return $processedContent;
             }
         } catch (Exception $e) {
-            Logger::getInstance()->log($cleanUrl, 'WAYBACK_FETCH_ERROR', $e->getMessage());
+            error_log("WAYBACK_FETCH_ERROR: " . $e->getMessage());
+        }
+
+        // 7. Tenta buscar com Selenium como fallback
+        try {
+            $content = $this->fetchFromSelenium($cleanUrl, 'firefox');
+            if (!empty($content)) {
+                $processedContent = $this->processContent($content, $host, $cleanUrl);
+                $this->cache->set($cleanUrl, $processedContent);
+                return $processedContent;
+            }
+        } catch (Exception $e) {
+            error_log("SELENIUM_ERROR: " . $e->getMessage());
         }
 
         Logger::getInstance()->log($cleanUrl, 'GENERAL_FETCH_ERROR');