Load by default as Html when content type is null or empty

zzzprojects · May 22, 2017 · 6584a09 · 6584a09
1 parent 3fcf79a
commit 6584a09
Showing 1 changed file with 21 additions and 0 deletions.
diff --git a/src/HtmlAgilityPack/HtmlWeb.cs b/src/HtmlAgilityPack/HtmlWeb.cs
@@ -1521,6 +1521,7 @@ private HttpStatusCode Get(Uri uri, string method, string path, HtmlDocument doc
             _responseUri = resp.ResponseUri;
 
             bool html = IsHtmlContent(resp.ContentType);
+            bool isUnknown = string.IsNullOrEmpty(resp.ContentType);
 
             Encoding respenc = !string.IsNullOrEmpty(resp.ContentEncoding)
                 ? Encoding.GetEncoding(resp.ContentEncoding)
@@ -1576,6 +1577,26 @@ private HttpStatusCode Get(Uri uri, string method, string path, HtmlDocument doc
                             doc.Load(s, respenc);
                         }
                     }
+
+                    if (doc != null && isUnknown)
+                    {
+                        try
+                        {
+                            if (respenc == null)
+                            {
+                                doc.Load(s, true);
+                            }
+                            else
+                            {
+                                doc.Load(s, respenc);
+                            }
+                        }
+                        catch
+                        {
+                            // Thats fine, the content type was unknown so probably not HTML
+                            // Perhaps trying to figure if the content contains some HTML before would be a better idea.
+                        }
+                    }
                 }
                 resp.Close();
             }