From 07b811019e75bc209be98b5d8133b4bb6d8541e6 Mon Sep 17 00:00:00 2001
From: Tao Sun <168447269+fengju0213@users.noreply.github.com>
Date: Tue, 22 Apr 2025 12:19:24 +0800
Subject: [PATCH] Update document_toolkit.py

---
 owl/utils/document_toolkit.py | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/owl/utils/document_toolkit.py b/owl/utils/document_toolkit.py
index deccda2..f13faee 100644
--- a/owl/utils/document_toolkit.py
+++ b/owl/utils/document_toolkit.py
@@ -117,8 +117,19 @@ class DocumentProcessingToolkit(BaseToolkit):
                 return True, content
 
         if self._is_webpage(document_path):
-            extracted_text = self._extract_webpage_content(document_path)
-            return True, extracted_text
+            try:
+                extracted_text = self._extract_webpage_content(document_path)
+                return True, extracted_text
+            except Exception:
+                try:
+                    elements = self.uio.parse_file_or_url(document_path)
+                    if elements is None:
+                        logger.error(f"Failed to parse the document: {document_path}.")
+                        return False, f"Failed to parse the document: {document_path}."
+                    else:
+                        return True, elements
+                except Exception:
+                    return False, "Failed to extract content from the webpage."
 
         else:
             try: