Update app.py

app.py

@@ -3,7 +3,7 @@ from openpyxl import load_workbook
 from pptx import Presentation
 import gradio as gr
 import io
-import docx2python
+import docx2python  
 from huggingface_hub import InferenceClient
 
 # Initialize the Mistral chat model
@@ -52,16 +52,10 @@ def read_document(file):
 
     elif file_extension == 'doc' or file_extension == 'docx':
         try:
-            doc_result = docx2python.convert(io.BytesIO(file_content))
+            doc_result = docx2python.Docx2Python(io.BytesIO(file_content)).process()  
             content = ''
-            for page in doc_result:
-                for paragraph in page:
-                    if isinstance(paragraph, str):
-                        content += paragraph + ' '
-                    elif isinstance(paragraph, list):
-                        for sub_paragraph in paragraph:
-                            if isinstance(sub_paragraph, str):
-                                content += sub_paragraph + ' '
+            for paragraph in doc_result.text: 
+                content += paragraph + ' '
             return content
         except Exception as e:
             return f"Error reading DOC/DOCX: {e}"