Update chat_app_remote.py

chat_app_remote.py

@@ -12,7 +12,6 @@ import scipy.io.wavfile as wavfile
 
 client = Client(os.environ['src'])
 
-
 def create_frontend_demo():
     def chat_function(message, history, session_id):
         if not session_id:
@@ -25,13 +24,34 @@ def create_frontend_demo():
             fn_index=0             
         )
         
-        # The backend returns: empty_string, history, audio_path, display_text
+
         _, new_history, audio_path, display_text = result
         
-        # For audio, we need to return the path string directly
+
         return "", new_history, audio_path, session_id, display_text
 
-    with gr.Blocks(theme="Respair/[email protected]") as demo:
+    with gr.Blocks(theme="Respair/[email protected]", css="""
+        .input-row {
+            display: flex !important;
+            gap: 0 !important;
+        }
+        .input-row > div {
+            flex: 1;
+        }
+        .input-row > div:last-child {
+            flex: 0 0 auto;
+            width: auto;
+        }
+        .input-row button {
+            border-top-left-radius: 0;
+            border-bottom-left-radius: 0;
+            margin: 0;
+        }
+        .input-row input {
+            border-top-right-radius: 0;
+            border-bottom-right-radius: 0;
+        }
+    """) as demo:
         session_id_state = gr.State("")
         
         with gr.Tabs() as tabs:
@@ -47,11 +67,13 @@ def create_frontend_demo():
                     placeholder="Start chatting with Aira..."
                 )
                 
-                msg = gr.Textbox(
-                    show_label=False,
-                    placeholder="Enter text and press enter",
-                    container=False
-                )
+                with gr.Row(elem_classes="input-row"):
+                    msg = gr.Textbox(
+                        show_label=False,
+                        placeholder="Enter text and press enter",
+                        container=False
+                    )
+                    send_btn = gr.Button("Send")
                 
                 audio_output = gr.Audio(
                     label="Aira's Response",
@@ -85,7 +107,8 @@ def create_frontend_demo():
                     You can talk to her in English or Japanese, but she will only respond in Japanese (Subs over dubs, bros) ask her to give you a Subtitle if you can't talk in Japanese. <br>
                     
                     The majority of the latency depends on the HF's inference api.
-                    LLM is not fine-tuned or optimized at all. the current state of conversational off-the-shelf japanese LLM seem to be less than remarkable, please beware of that.
+                    LLM is not fine-tuned. 
+                    the current state of conversational off-the-shelf japanese LLM seem to be less than remarkable, please beware of that.
                     
                     1. Enter your Session ID above or leave blank for a new one
                     2. Click 'Set Session ID' to confirm
@@ -105,6 +128,12 @@ def create_frontend_demo():
             outputs=[msg, chatbot, audio_output, session_id_state, session_display]
         )
 
+        send_btn.click(
+            respond,
+            inputs=[msg, chatbot, session_id_state],
+            outputs=[msg, chatbot, audio_output, session_id_state, session_display]
+        )
+
         def set_session(user_id):
             result = client.predict(
                 user_id,
@@ -131,7 +160,7 @@ def create_frontend_demo():
                     
                     audio = {"path": temp.name, "meta": {"_type": "gradio.FileData"}}
 
-                    # Get the result while the temporary file still exists
+
                     result = client.predict(
                         audio,
                         history,
@@ -139,7 +168,7 @@ def create_frontend_demo():
                         api_name="/handle_audio"
                     )
                     
-                    # Unpack only 3 values and construct the display text
+            
                     audio_path, new_history, new_session_id = result
                     display_text = f"Current Session ID: {new_session_id}"
                 
@@ -148,7 +177,7 @@ def create_frontend_demo():
             except Exception as e:
                 print(f"Error processing audio: {str(e)}")
                 import traceback
-                traceback.print_exc()  # This will print the full error traceback
+                traceback.print_exc()  
                 return None, history, session_id, f"Error processing audio. Session ID: {session_id}"
             
         audio_input.stop_recording(