ver Mar11th

updated filter_nodes

mm_agents/gpt_4v_agent.py

@@ -8,6 +8,7 @@ import uuid
 from http import HTTPStatus
 from io import BytesIO
 from typing import Dict, List
+import xml.etree.ElementTree as ET
 
 import backoff
 import dashscope
@@ -40,8 +41,8 @@ def encode_image(image_path):
 
 
 def linearize_accessibility_tree(accessibility_tree):
-    leaf_nodes = find_leaf_nodes(accessibility_tree)
-    filtered_nodes = filter_nodes(leaf_nodes)
+    #leaf_nodes = find_leaf_nodes(accessibility_tree)
+    filtered_nodes = filter_nodes(ET.fromstring(accessibility_tree))
 
     linearized_accessibility_tree = "tag\tname\ttext\tposition\tsize\n"
     # Linearize the accessibility tree nodes into a table format