ajustes e melhorias

app/inc/URLAnalyzer.php

@@ -110,13 +110,13 @@ class URLAnalyzer
             return $this->cache->get($cleanUrl);
         }
 
-        $domain = parse_url($cleanUrl, PHP_URL_HOST);
+        $host = parse_url($cleanUrl, PHP_URL_HOST);
-        $domain = preg_replace('/^www\./', '', $domain);
+        $host = preg_replace('/^www\./', '', $host);
 
         // Verificação de domínios bloqueados
         foreach (BLOCKED_DOMAINS as $blockedDomain) {
             // Verifica apenas correspondência exata do domínio
-            if ($domain === $blockedDomain) {
+            if ($host === $blockedDomain) {
                 $error = 'Este domínio está bloqueado para extração.';
                 $this->logError($cleanUrl, $error);
                 throw new Exception($error);
@@ -127,10 +127,11 @@ class URLAnalyzer
 
         // Primeiro, tenta buscar o conteúdo diretamente
         try {
-            $content = $this->fetchContent($url);
+            $content = $this->fetchContent($url, $host);
         } catch (Exception $e) {
             // Se falhar, registra o erro de busca direta
             $this->logError($url, "Direct fetch error: " . $e->getMessage());
+            $content = null;
         }
 
         // Se a busca direta falhar, tenta o Wayback Machine
@@ -144,12 +145,11 @@ class URLAnalyzer
         }
 
         if (!empty($content)) {
-            $content = $this->processContent($content, $domain, $cleanUrl);
+            $content = $this->processContent($content, $host, $cleanUrl);
             $this->cache->set($cleanUrl, $content);
-            return $content;
         }
         
-        return null;
+        return $content;
     }
 
     /**
@@ -212,15 +212,11 @@ class URLAnalyzer
      * Realiza requisição HTTP usando Curl Class
      * 
      * @param string $url URL para requisição
-     * @return string Conteúdo obtido
+     * @return array Resultado da requisição com conteúdo e código HTTP
-     * @throws Exception Em caso de erro na requisição
      */
-    private function fetchContent($url)
+    private function fetchContent($url, $host)
     {
-        $parsedUrl = parse_url($url);
+        $domainRules = $this->getDomainRules($host);
-        $host = $parsedUrl['host'];
-
-        $domainRules = $this->getDomainRules(parse_url($url, PHP_URL_HOST));
 
         // Obtém a configuração do user agent
         $curl = new Curl();
@@ -323,78 +319,6 @@ class URLAnalyzer
         return $this->rules->getDomainRules($domain);
     }
 
-    /**
-     * Remove classes específicas de um elemento
-     * 
-     * @param DOMElement $element Elemento DOM
-     * @param array $classesToRemove Classes a serem removidas
-     */
-    private function removeClassNames($element, $classesToRemove)
-    {
-        if (!$element->hasAttribute('class')) {
-            return;
-        }
-
-        $classes = explode(' ', $element->getAttribute('class'));
-        $newClasses = array_filter($classes, function ($class) use ($classesToRemove) {
-            return !in_array(trim($class), $classesToRemove);
-        });
-
-        if (empty($newClasses)) {
-            $element->removeAttribute('class');
-        } else {
-            $element->setAttribute('class', implode(' ', $newClasses));
-        }
-    }
-
-    /**
-     * Corrige URLs relativas em um documento DOM
-     * 
-     * @param DOMDocument $dom Documento DOM
-     * @param DOMXPath $xpath Objeto XPath
-     * @param string $baseUrl URL base para correção
-     */
-    private function fixRelativeUrls($dom, $xpath, $baseUrl)
-    {
-        $parsedBase = parse_url($baseUrl);
-        $baseHost = $parsedBase['scheme'] . '://' . $parsedBase['host'];
-
-        $elements = $xpath->query("//*[@src]");
-        if ($elements !== false) {
-            foreach ($elements as $element) {
-                if ($element instanceof DOMElement) {
-                    $src = $element->getAttribute('src');
-                    if (strpos($src, 'base64') !== false) {
-                        continue;
-                    }
-                    if (strpos($src, 'http') !== 0 && strpos($src, '//') !== 0) {
-                        $src = ltrim($src, '/');
-                        $element->setAttribute('src', $baseHost . '/' . $src);
-                    }
-                }
-            }
-        }
-
-        $elements = $xpath->query("//*[@href]");
-        if ($elements !== false) {
-            foreach ($elements as $element) {
-                if ($element instanceof DOMElement) {
-                    $href = $element->getAttribute('href');
-                    if (strpos($href, 'mailto:') === 0 || 
-                        strpos($href, 'tel:') === 0 || 
-                        strpos($href, 'javascript:') === 0 || 
-                        strpos($href, '#') === 0) {
-                        continue;
-                    }
-                    if (strpos($href, 'http') !== 0 && strpos($href, '//') !== 0) {
-                        $href = ltrim($href, '/');
-                        $element->setAttribute('href', $baseHost . '/' . $href);
-                    }
-                }
-            }
-        }
-    }
-
     /**
      * Processa o conteúdo HTML aplicando regras do domínio
      * 
@@ -403,7 +327,7 @@ class URLAnalyzer
      * @param string $url URL completa
      * @return string Conteúdo processado
      */
-    private function processContent($content, $domain, $url)
+    private function processContent($content, $host, $url)
     {
         $dom = new DOMDocument();
         $dom->preserveWhiteSpace = true;
@@ -435,19 +359,10 @@ class URLAnalyzer
         // Sempre aplica a correção de URLs relativas
         $this->fixRelativeUrls($dom, $xpath, $url);
 
-        $domainRules = $this->getDomainRules($domain);
+        $domainRules = $this->getDomainRules($host);
-        if ($domainRules !== null) {
         if (isset($domainRules['customStyle'])) {
             $styleElement = $dom->createElement('style');
-                $styleContent = '';
+            $styleElement->appendChild($dom->createTextNode($domainRules['customStyle']));
-                foreach ($domainRules['customStyle'] as $selector => $rules) {
-                    if (is_array($rules)) {
-                        $styleContent .= $selector . ' { ' . implode('; ', $rules) . ' } ';
-                    } else {
-                        $styleContent .= $selector . ' { ' . $rules . ' } ';
-                    }
-                }
-                $styleElement->appendChild($dom->createTextNode($styleContent));
             $dom->getElementsByTagName('head')[0]->appendChild($styleElement);
         }
 
@@ -518,7 +433,6 @@ class URLAnalyzer
                 }
             }
         }
-        }
 
         $elements = $xpath->query("//*[@style]");
         if ($elements !== false) {
@@ -535,7 +449,7 @@ class URLAnalyzer
         $body = $xpath->query('//body')->item(0);
         if ($body) {
             $marretaDiv = $dom->createElement('div');
-            $marretaDiv->setAttribute('style', 'z-index: 99999; position: fixed; bottom: 0; right: 4px; background: rgb(37,99,235); color: #fff; font-size: 13px; line-height: 1em; padding: 6px; margin: 0px; overflow: hidden; border-top-left-radius: 3px; border-top-right-radius: 3px; font-family: Tahoma, sans-serif;');
+            $marretaDiv->setAttribute('style', 'z-index: 99999; position: fixed; top: 0; right: 4px; background: rgb(37,99,235); color: #fff; font-size: 13px; line-height: 1em; padding: 6px; margin: 0px; overflow: hidden; border-bottom-left-radius: 3px; border-bottom-right-radius: 3px; font-family: Tahoma, sans-serif;');
             $marretaHtml = $dom->createDocumentFragment();
             $marretaHtml->appendXML('Chapéu de paywall é <a href="'.SITE_URL.'" style="color: #fff; text-decoration: underline; font-weight: bold;" target="_blank">Marreta</a>!');
             $marretaDiv->appendChild($marretaHtml);
@@ -544,4 +458,76 @@ class URLAnalyzer
 
         return $dom->saveHTML();
     }
+
+    /**
+     * Remove classes específicas de um elemento
+     * 
+     * @param DOMElement $element Elemento DOM
+     * @param array $classesToRemove Classes a serem removidas
+     */
+    private function removeClassNames($element, $classesToRemove)
+    {
+        if (!$element->hasAttribute('class')) {
+            return;
+        }
+
+        $classes = explode(' ', $element->getAttribute('class'));
+        $newClasses = array_filter($classes, function ($class) use ($classesToRemove) {
+            return !in_array(trim($class), $classesToRemove);
+        });
+
+        if (empty($newClasses)) {
+            $element->removeAttribute('class');
+        } else {
+            $element->setAttribute('class', implode(' ', $newClasses));
+        }
+    }
+
+    /**
+     * Corrige URLs relativas em um documento DOM
+     * 
+     * @param DOMDocument $dom Documento DOM
+     * @param DOMXPath $xpath Objeto XPath
+     * @param string $baseUrl URL base para correção
+     */
+    private function fixRelativeUrls($dom, $xpath, $baseUrl)
+    {
+        $parsedBase = parse_url($baseUrl);
+        $baseHost = $parsedBase['scheme'] . '://' . $parsedBase['host'];
+
+        $elements = $xpath->query("//*[@src]");
+        if ($elements !== false) {
+            foreach ($elements as $element) {
+                if ($element instanceof DOMElement) {
+                    $src = $element->getAttribute('src');
+                    if (strpos($src, 'base64') !== false) {
+                        continue;
+                    }
+                    if (strpos($src, 'http') !== 0 && strpos($src, '//') !== 0) {
+                        $src = ltrim($src, '/');
+                        $element->setAttribute('src', $baseHost . '/' . $src);
+                    }
+                }
+            }
+        }
+
+        $elements = $xpath->query("//*[@href]");
+        if ($elements !== false) {
+            foreach ($elements as $element) {
+                if ($element instanceof DOMElement) {
+                    $href = $element->getAttribute('href');
+                    if (strpos($href, 'mailto:') === 0 || 
+                        strpos($href, 'tel:') === 0 || 
+                        strpos($href, 'javascript:') === 0 || 
+                        strpos($href, '#') === 0) {
+                        continue;
+                    }
+                    if (strpos($href, 'http') !== 0 && strpos($href, '//') !== 0) {
+                        $href = ltrim($href, '/');
+                        $element->setAttribute('href', $baseHost . '/' . $href);
+                    }
+                }
+            }
+        }
+    }
 }