regra de corrigir paginas sem urls relativas se torna global

Author: altendorfme

app/data/domain_rules.php

@@ -14,7 +14,6 @@
  * - scriptTagRemove: Array de scripts que devem ser removidos (partial match)
  * - cookies: Array associativo de cookies a serem definidos (null remove o cookie)
  * - classAttrRemove: Array de classes a serem removidas de elementos
- * - fixRelativeUrls: Boolean para habilitar correção de URLs relativas
  * - customCode: String contendo código JavaScript personalizado para execução
  * - excludeGlobalRules: Array de regras globais que devem ser ignoradas
  */
@@ -51,24 +50,8 @@
             'estadao_paywall' => null
         ]
     ],
-    'exame.com' => [
-        'fixRelativeUrls' => true,
-    ],
-    'diarinho.net' => [
-        'fixRelativeUrls' => true,
-    ],
-    'em.com.br' => [
-        'fixRelativeUrls' => true,
-    ],
-    'businessinsider.com' => [
-        'fixRelativeUrls' => true,
-    ],
     'opovo.com.br' => [
-        'fixRelativeUrls' => true,
-        'classElementRemove' => ['screen-loading', 'overlay-advise'],
-    ],
-    'folhadelondrina.com.br' => [
-        'fixRelativeUrls' => true,
+        'classElementRemove' => ['screen-loading', 'overlay-advise']
     ],
     'crusoe.com.br' => [
         'cookies' => [
@@ -125,7 +108,7 @@ function checkPaywall() {
     ],
     'foreignpolicy.com' => [
         'idElementRemove' => ['paywall_bg'],
-        'classAttrRemove' => ['overlay-no-scroll', 'overlay-no-scroll'],
+        'classAttrRemove' => ['overlay-no-scroll', 'overlay-no-scroll']
     ],
     'dgabc.com.br' => [
         'customCode' => '
@@ -135,8 +118,7 @@ function checkPaywall() {
                 $(".linhaSuperBanner").show();
                 $(".footer").show();
                 $(".NoticiaExclusivaLogado").show();
-            ',
-        'fixRelativeUrls' => true,
+            '
     ],
     'forbes.com' => [
         'classElementRemove' => ['zephr-backdrop', 'zephr-generic-modal'],
@@ -149,7 +131,7 @@ function checkPaywall() {
         ],
     ],
     'seudinheiro.com' => [
-        'idElementRemove' => ['premium-paywall'],
+        'idElementRemove' => ['premium-paywall']
     ],
     'technologyreview.com' => [
         'cookies' => [

app/data/global_rules.php

@@ -54,6 +54,7 @@
     ],
     'scriptTagRemove' => [
         'tracking' => [
+            'gtm.js',
             'ga.js',
             'fbevents.js',
             'pixel.js',

app/inc/URLAnalyzer.php

@@ -611,6 +611,9 @@ private function fixRelativeUrls($dom, $xpath, $baseUrl)
             foreach ($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 . '/' . $src);
@@ -624,7 +627,13 @@ private function fixRelativeUrls($dom, $xpath, $baseUrl)
             foreach ($elements as $element) {
                 if ($element instanceof DOMElement) {
                     $href = $element->getAttribute('href');
-                    if (strpos($href, 'http') !== 0 && strpos($href, '//') !== 0 && strpos($href, '#') !== 0 && strpos($href, 'mailto:') !== 0) {
+                    if (strpos($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 . '/' . $href);
                     }
@@ -651,12 +660,11 @@ private function processContent($content, $domain, $url)
 
         $xpath = new DOMXPath($dom);
 
+        // Sempre aplica a correção de URLs relativas
+        $this->fixRelativeUrls($dom, $xpath, $url);
+
         $domainRules = $this->getDomainRules($domain);
         if ($domainRules !== null) {
-            if (isset($domainRules['fixRelativeUrls']) && $domainRules['fixRelativeUrls'] === true) {
-                $this->fixRelativeUrls($dom, $xpath, $url);
-            }
-
             if (isset($domainRules['customStyle'])) {
                 $styleElement = $dom->createElement('style');
                 $styleContent = '';