Fix #149

[github/wallabag/wallabag.git] / inc / poche / Url.class.php
diff --git a/inc/poche/Url.class.php b/inc/poche/Url.class.php

index f4a8f99e6377dee187142db7495227f8a49860c1..77ed53985ac336719999082977d40d4c5cd1cfbc 100644 (file)
--- a/inc/poche/Url.class.php
+++ b/inc/poche/Url.class.php
@@ -27,7 +27,7 @@ class Url
  
      public function isCorrect()
      {
-        $pattern = '|^http(s)?://[a-z0-9-]+(.[a-z0-9-]+)*(:[0-9]+)?(/.*)?$|i';
+        $pattern = '|^(.*:)//([a-z\-.]+)(:[0-9]+)?(.*)$|i';
  
          return preg_match($pattern, $this->url);
      }
@@ -64,8 +64,15 @@ class Url
              if (function_exists('tidy_parse_string')) {
                  $tidy = tidy_parse_string($html, array(), 'UTF8');
                  $tidy->cleanRepair();
-                $html = $tidy->value;
-            }
+
+                               //Warning: tidy might fail so, ensure there is still a content
+                               $body = $tidy->body();
+
+                               //hasChildren does not seem to work, just check the string
+                               //returned (and do not forget to clean the white spaces)
+                               if (preg_replace('/\s+/', '', $body->value) !== "<body></body>")
+                                       $html = $tidy->value;
+            } 
  
              $parameters = array();
              if (isset($html) and strlen($html) > 0)