prevent to throw exceptions if the html is not parsed correcly #58

Author: oscarotero

src/Providers/Html.php

@@ -29,9 +29,9 @@ public function run()
         self::extractFromLink($html, $this->bag);
         self::extractFromMeta($html, $this->bag);
 
-        $main = self::getMainElement($html);
-
-        self::extractImages($main, $this->bag, $this->request->getDomain());
+        if ($main = self::getMainElement($html)) {
+            self::extractImages($main, $this->bag, $this->request->getDomain());
+        }
 
         //Title
         $title = $html->getElementsByTagName('title');

src/Request.php

@@ -190,6 +190,8 @@ public function getHtmlContent()
                 }
 
                 $errors = libxml_use_internal_errors(true);
+                $entities = libxml_disable_entity_loader(true);
+
                 $this->htmlContent = new \DOMDocument();
 
                 if ((mb_detect_encoding($content) === 'UTF-8') && mb_check_encoding($content, 'UTF-8')) {
@@ -198,7 +200,9 @@ public function getHtmlContent()
                 }
 
                 $this->htmlContent->loadHTML($content);
+
                 libxml_use_internal_errors($errors);
+                libxml_disable_entity_loader($entities);
             } catch (\Exception $E) {
                 return $this->htmlContent = false;
             }

src/Url.php

@@ -516,6 +516,7 @@ private function setPath($path)
 
         if ($this->getScheme() === 'data') {
             $this->info['content'] = $path;
+
             return;
         }