refactor: render from output and derive content at save

Author: Algorithm5838

backend/open_webui/utils/middleware.py

@@ -3422,13 +3422,20 @@ async def outlet_filter_handler(ctx):
                     outlet_message_id = message.get('id')
                     if outlet_message_id and outlet_message_id in messages_map:
                         original_message = messages_map[outlet_message_id]
-                        if original_message.get('content') != message.get('content'):
+                        content_changed = original_message.get('content') != message.get('content')
+                        output_changed = message.get('output') and message.get('output') != original_message.get('output')
+                        if content_changed or output_changed:
+                            # If output was modified, re-derive content from it
+                            new_content = message.get('content', original_message.get('content', ''))
+                            if output_changed:
+                                new_content = serialize_output(message['output'])
                             await Chats.upsert_message_to_chat_by_id_and_message_id(
                                 chat_id,
                                 outlet_message_id,
                                 {
-                                    'content': message['content'],
+                                    'content': new_content,
                                     'originalContent': original_message.get('content'),
+                                    **({'output': message['output']} if output_changed else {}),
                                 },
                             )
 
@@ -4069,7 +4076,6 @@ async def flush_pending_delta_data(threshold: int = 0):
 
                                     processed_data = {
                                         'output': full_output(),
-                                        'content': serialize_output(full_output()),
                                     }
 
                                     # print(data)

src/lib/components/chat/Chat.svelte

@@ -1805,6 +1805,12 @@
 			message.output = output;
 		}
 
+		// When the response is done and content is provided, set it for copy/TTS
+		// (the frontend renders from output during streaming, content is the final cache)
+		if (done && content) {
+			message.content = content;
+		}
+
 		if (error) {
 			await handleOpenAIError(error, message);
 		}
@@ -1813,7 +1819,9 @@
 			message.sources = sources;
 		}
 
-		if (choices) {
+		// Only accumulate content from choices when there's no structured output
+		// (output-based rendering takes priority — avoids redundant serialize/parse round-trip)
+		if (choices && !output) {
 			if (choices[0]?.message?.content) {
 				// Non-stream response
 				message.content += choices[0]?.message?.content;