Update app.py

app.py

@@ -199,9 +199,10 @@ def process_urls(url_input, bulk_toggle, action_radio, max_urls, crawl_depth, pr
         
         # Prepare output data
         if action_radio in ['Scrape data', 'Both']:
+            cleaned_content = BeautifulSoup(latest_html, 'html.parser').get_text(separator="\n").strip()
             scraped_data.append({
                 'url': url,
-                'content': latest_html,
+                'content': cleaned_content,
                 'timestamp': datetime.datetime.now().isoformat(),
                 'changes_detected': changes_log
             })
@@ -231,19 +232,20 @@ def process_urls(url_input, bulk_toggle, action_radio, max_urls, crawl_depth, pr
                 }
                 zipf.writestr('data.json', json.dumps(data_to_save, indent=2))
         
-        # Get the path to the temporary file
+        # # Get the path to the temporary file
         zip_file_path = tmp_file.name
     
     # Prepare display data
     display_data = {
         'scraped_urls': len(scraped_data),
         'screenshots_taken': len(screenshots),
-        'changes_detected': changes_log
+        'changes_detected': changes_log,
+        'screenshots': [(screenshot_url, screenshot_data) for screenshot_url, screenshot_data in screenshots]
     }
     
-    # Return the path to the temporary ZIP file and display data
-    return zip_file_path, json.dumps(display_data, indent=2)
-    
+    # Return the path to the temporary ZIP file, display data, and screenshots
+    return zip_file_path, json.dumps(display_data, indent=2), [screenshot for _, screenshot in screenshots]
+
 def create_interface():
     """Create the Gradio interface."""
     with gr.Blocks(theme=gr.themes.Soft()) as demo:
@@ -289,8 +291,9 @@ def create_interface():
                 process_button = gr.Button("Process URLs", variant="primary")
                 
                 with gr.Column():
-                    screenshot_zip = gr.File(label="Download Results")  # Removed file_name
+                    screenshot_zip = gr.File(label="Download Results")
                     scraped_data_output = gr.JSON(label="Results Summary")
+                    screenshot_gallery = gr.Gallery(label="Screenshots", show_label=True).style(grid=2)
                 
                 process_button.click(
                     fn=process_urls,
@@ -303,7 +306,8 @@ def create_interface():
                     ],
                     outputs=[
                         screenshot_zip,
-                        scraped_data_output
+                        scraped_data_output,
+                        screenshot_gallery
                     ],
                     show_progress=True
                 )