Don't depend on protocol in video regex

2014-01-11 10:00:00 +01:00
parent 466518d57c
commit 845b1438fb
2 changed files with 13 additions and 11 deletions
@@ -111,7 +111,7 @@ namespace ReadSharp.Ports.NReadability
    private static readonly Regex _BreakBeforeParagraphRegex = new Regex("<br[^>]*>\\s*<p");
    private static readonly Regex _NormalizeSpacesRegex = new Regex("\\s{2,}");
    private static readonly Regex _KillBreaksRegex = new Regex("(<br\\s*\\/?>(\\s|&nbsp;?)*){1,}");
-    private static readonly Regex _VideoRegex = new Regex("http:\\/\\/(www\\.)?(youtube|vimeo)\\.com", RegexOptions.IgnoreCase);
+    private static readonly Regex _VideoRegex = new Regex("\\/\\/(www\\.)?(youtube|vimeo)\\.com", RegexOptions.IgnoreCase);
    private static readonly Regex _ReplaceDoubleBrsRegex = new Regex("(<br[^>]*>[ \\n\\r\\t]*){2,}", RegexOptions.IgnoreCase);
    private static readonly Regex _ReplaceFontsRegex = new Regex("<(\\/?)font[^>]*>", RegexOptions.IgnoreCase);
    private static readonly Regex _ArticleTitleDashRegex1 = new Regex(" [\\|\\-] ");
@@ -1435,6 +1435,8 @@ namespace ReadSharp.Ports.NReadability

      var elementsToRemove = new List<XElement>();

+      // TODO: proof iframe videos
+
      foreach (XElement element in elements)
      {
        /* Allow youtube and vimeo videos through as people usually want to see those. */