Update app.py

app.py

@@ -290,7 +290,6 @@ def convert_chat_messages_to_gradio_format(messages):
     """
     gradio_chat = []
     user_text, assistant_text = None, None
-
     for msg in messages:
         if msg.role == "user":
             if user_text is not None or assistant_text is not None:
@@ -304,10 +303,8 @@ def convert_chat_messages_to_gradio_format(messages):
                 assistant_text = msg.content
             else:
                 assistant_text += msg.content
-
     if user_text is not None or assistant_text is not None:
         gradio_chat.append((user_text or "", assistant_text or ""))
-
     return gradio_chat
 
 def stream_gemini_response(user_message: str, messages: list) -> Iterator[list]:
@@ -316,17 +313,14 @@ def stream_gemini_response(user_message: str, messages: list) -> Iterator[list]:
     """
     if not user_message.strip():
         user_message = "...(No content from user)..."
-
     try:
         print(f"\n=== [Gemini] New Request ===\nUser message: '{user_message}'")
         chat_history = format_chat_history(messages)
         chat = model.start_chat(history=chat_history)
         response = chat.send_message(user_message, stream=True)
-
         thought_buffer = ""
         response_buffer = ""
         thinking_complete = False
-
         # "Thinking" 역할 추가
         messages.append(
             ChatMessage(
@@ -336,11 +330,9 @@ def stream_gemini_response(user_message: str, messages: list) -> Iterator[list]:
             )
         )
         yield convert_chat_messages_to_gradio_format(messages)
-
         for chunk in response:
             parts = chunk.candidates[0].content.parts
             current_chunk = parts[0].text
-
             if len(parts) == 2 and not thinking_complete:
                 thought_buffer += current_chunk
                 messages[-1] = ChatMessage(
@@ -349,7 +341,6 @@ def stream_gemini_response(user_message: str, messages: list) -> Iterator[list]:
                     metadata={"title": "⚙️ Thinking: *The thoughts produced by the model are experimental"}
                 )
                 yield convert_chat_messages_to_gradio_format(messages)
-
                 response_buffer = parts[1].text
                 messages.append(ChatMessage(role="assistant", content=response_buffer))
                 thinking_complete = True
@@ -363,11 +354,8 @@ def stream_gemini_response(user_message: str, messages: list) -> Iterator[list]:
                     content=thought_buffer,
                     metadata={"title": "⚙️ Thinking: *The thoughts produced by the model are experimental"}
                 )
-
             yield convert_chat_messages_to_gradio_format(messages)
-
         print(f"\n=== [Gemini] Final Response ===\n{response_buffer}")
-
     except Exception as e:
         print(f"\n=== [Gemini] Error ===\n{str(e)}")
         messages.append(ChatMessage(role="assistant", content=f"I encountered an error: {str(e)}"))
@@ -381,15 +369,14 @@ def user_message(msg: str, history: list, doc_text: str) -> tuple[str, list]:
         user_query = f"다음 문서를 참고하여 답변:\n\n{doc_text}\n\n질문: {msg}"
     else:
         user_query = msg
-
     history.append(ChatMessage(role="user", content=user_query))
     return "", history
 
 def reset_states(file_a, file_b):
     """
-    새 파일 업로드 시 chat_history와 md_state를 초기화 (Chatbot은 숨김 처리)
+    새 파일 업로드 시 chat_history와 md_state를 초기화
     """
-    return [], "", []
+    return [], "", ""
 
 ###############################
 # UI 통합
@@ -405,27 +392,29 @@ if __name__ == "__main__":
             </div>
             """)
             
-            md_state = gr.State("")      # 변환된 마크다운 텍스트 (두 파일의 결과 통합)
+            # 변환 결과를 보여줄 visible Markdown 컴포넌트
+            conversion_md = gr.Markdown(label="변환 결과", visible=True)
+            md_state = gr.State("")      # 내부 상태 (문서 변환 결과)
             chat_history = gr.State([])  # ChatMessage 리스트
 
-            # 단일(숨김) Chatbot 컴포넌트: 화면에 보이지 않음
-            hidden_chatbot = gr.Chatbot(visible=False)
+            # visible Chatbot 컴포넌트
+            chatbot = gr.Chatbot(visible=True)
 
             with gr.Row():
                 file_a = gr.File(label="문서 A 업로드", file_types=[".pdf", ".png", ".jpeg", ".jpg", ".txt", ".csv"], interactive=True)
                 file_b = gr.File(label="문서 B 업로드", file_types=[".pdf", ".png", ".jpeg", ".jpg", ".txt", ".csv"], interactive=True)
                 convert_btn = gr.Button("비교용 변환하기")
 
-            # 파일 업로드 시 상태 초기화 (Chatbot은 숨김 처리)
+            # 파일 업로드 시 상태 초기화
             file_a.change(
                 fn=reset_states,
                 inputs=[file_a, file_b],
-                outputs=[chat_history, md_state, hidden_chatbot]
+                outputs=[chat_history, md_state, chatbot]
             )
             file_b.change(
                 fn=reset_states,
                 inputs=[file_a, file_b],
-                outputs=[chat_history, md_state, hidden_chatbot]
+                outputs=[chat_history, md_state, chatbot]
             )
 
             max_pages = gr.Slider(1, 20, 10, visible=False)
@@ -438,7 +427,7 @@ if __name__ == "__main__":
             convert_btn.click(
                 fn=to_markdown_comparison,
                 inputs=[file_a, file_b, max_pages, is_ocr, layout_mode, formula_enable, table_enable, language],
-                outputs=md_state,
+                outputs=conversion_md,
                 show_progress=True
             )
 
@@ -455,24 +444,23 @@ if __name__ == "__main__":
                 chat_input = gr.Textbox(lines=1, placeholder="질문을 입력하세요...")
                 clear_btn = gr.Button("대화 초기화")
 
-            # 채팅 입력 후 LLM 응답은 hidden_chatbot에 저장(화면에 보이지 않음)
             chat_input.submit(
                 fn=user_message,
-                inputs=[chat_input, chat_history, md_state],
+                inputs=[chat_input, chat_history, conversion_md],
                 outputs=[chat_input, chat_history]
             ).then(
                 fn=stream_gemini_response,
                 inputs=[chat_input, chat_history],
-                outputs=hidden_chatbot
+                outputs=chatbot
             )
 
             def clear_all():
-                return [], "", []
+                return [], "", ""
 
             clear_btn.click(
                 fn=clear_all,
                 inputs=[],
-                outputs=[chat_history, md_state, hidden_chatbot]
+                outputs=[chat_history, md_state, chatbot]
             )
 
     demo.launch(server_name="0.0.0.0", server_port=7860, debug=True, ssr_mode=True)