Update app.py

app.py

@@ -8,9 +8,6 @@ import google.generativeai as genai
 from io import BytesIO
 import PyPDF2
 from streamlit_webrtc import webrtc_streamer, WebRtcMode, RTCConfiguration
-import asyncio
-import numpy as np
-import pydub
 
 # Set your API key
 api_key = "AIzaSyAHD0FwX-Ds6Y3eI-i5Oz7IdbJqR6rN7pg"  # Replace with your actual API key
@@ -47,14 +44,6 @@ if 'chat_history' not in st.session_state:
     st.session_state['chat_history'] = []
 if 'file_uploader_key' not in st.session_state:
     st.session_state['file_uploader_key'] = str(uuid.uuid4())
-if 'uploaded_files' not in st.session_state:
-    st.session_state['uploaded_files'] = []
-if 'user_input' not in st.session_state:
-    st.session_state['user_input'] = ''
-if 'audio_data' not in st.session_state:
-    st.session_state['audio_data'] = None
-if 'is_recording' not in st.session_state:
-    st.session_state['is_recording'] = False
 
 # --- Streamlit UI ---
 st.title("Gemini Chatbot")
@@ -74,26 +63,22 @@ def get_file_base64(file_content, mime_type):
 def clear_conversation():
     st.session_state['chat_history'] = []
     st.session_state['file_uploader_key'] = str(uuid.uuid4())
-    st.session_state['user_input'] = ''
-    st.session_state['uploaded_files'] = []
-    st.session_state['audio_data'] = None
 
 def display_chat_history():
     chat_container = st.empty()
     with chat_container.container():
         for entry in st.session_state['chat_history']:
             role = entry["role"]
-            parts = entry["parts"]
-
+            parts = entry["parts"][0]
             if 'text' in parts:
-                st.markdown(f"{role.title()}: {parts['text']}")
+                st.markdown(f"**{role.title()}:** {parts['text']}")
             elif 'data' in parts:
                 mime_type = parts.get('mime_type', '')
                 if mime_type.startswith('image'):
                     st.image(Image.open(io.BytesIO(base64.b64decode(parts['data']))),
-                              caption='Uploaded Image', use_column_width=True)
+                             caption='Uploaded Image', use_column_width=True)
                 elif mime_type == 'application/pdf':
-                    st.write("PDF Content:")
+                    st.write("**PDF Content:**")
                     pdf_reader = PyPDF2.PdfReader(io.BytesIO(base64.b64decode(parts['data'])))
                     for page_num in range(len(pdf_reader.pages)):
                         page = pdf_reader.pages[page_num]
@@ -153,6 +138,7 @@ def send_message(audio_data=None):
     except Exception as e:
         st.error(f"An error occurred: {e}")
 
+    st.session_state.user_input = ''
     st.session_state.uploaded_files = []
     st.session_state.file_uploader_key = str(uuid.uuid4())
 
@@ -165,8 +151,7 @@ with col1:
     user_input = st.text_area(
         "Enter your message:",
         value="",
-        key="user_input",
-        on_change=lambda: setattr(st.session_state, 'user_input', '')
+        key="user_input"
     )
 with col2:
     send_button = st.button(
@@ -186,21 +171,10 @@ uploaded_files = st.file_uploader(
 # --- WebRTC Audio Recording ---
 RTC_CONFIGURATION = RTCConfiguration({"iceServers": [{"urls": ["stun:stun.l.google.com:19302"]}]})
 
-def convert_frames_to_wav(audio_frames):
-    """Convert audio frames to WAV format using PyDub."""
-    audio = np.concatenate(audio_frames)
-    audio_segment = pydub.AudioSegment(
-        audio.tobytes(),
-        frame_rate=16000,
-        sample_width=audio.dtype.itemsize,
-        channels=1
-    )
-    wav_io = BytesIO()
-    audio_segment.export(wav_io, format="wav")
-    wav_io.seek(0)
-    return wav_io.getvalue()
+if "webrtc_initialized" not in st.session_state:
+    st.session_state.webrtc_initialized = False
 
-async def run_webrtc():
+if not st.session_state.webrtc_initialized:
     webrtc_ctx = webrtc_streamer(
         key="audio-recorder",
         mode=WebRtcMode.SENDONLY,
@@ -208,17 +182,15 @@ async def run_webrtc():
         audio_receiver_size=256,
         media_stream_constraints={"video": False, "audio": True},
     )
-    if webrtc_ctx.audio_receiver:
-        st.session_state.is_recording = True
-        st.write("Recording audio...")
-        audio_frames = webrtc_ctx.audio_receiver.get_frames(timeout=None)
-        audio_data = [frame.to_ndarray() for frame in audio_frames]
-        st.session_state.audio_data = convert_frames_to_wav(audio_data)
+    st.session_state.webrtc_initialized = True
+    st.experimental_rerun() # Force Streamlit to rerun
 
+if webrtc_ctx.audio_receiver:
+    st.write("Recording audio...")
+    audio_frames = webrtc_ctx.audio_receiver.get_frames(timeout=None)
+    audio_data = b"".join([frame for frame in audio_frames])
     if st.button("Send Recording"):
-        send_message(audio_data=st.session_state.audio_data)
-    else:
-        st.session_state.is_recording = False
+        send_message(audio_data=audio_data)
 
 # --- Other Buttons ---
 st.button("Clear Conversation", on_click=clear_conversation)
@@ -230,16 +202,28 @@ st.session_state.uploaded_files = uploaded_files
 st.markdown(
     """
     <script>
-        document.addEventListener("keydown", function(event) {
-            if (event.ctrlKey && event.key === 'Enter') {
-                document.querySelector('button[type="primary"]').click();
+    document.addEventListener('DOMContentLoaded', (event) => {
+        document.querySelector('.stTextArea textarea').addEventListener('keydown', function(e) {
+            if (e.key === 'Enter' && e.ctrlKey) {
+                document.querySelector('.stButton > button').click();
+                e.preventDefault();
             }
         });
+    });
     </script>
     """,
     unsafe_allow_html=True
 )
 
-# --- Run WebRTC and display chat history ---
-asyncio.run(run_webrtc())
-display_chat_history()
+# --- Display Chat History ---
+display_chat_history()
+content_copy
+Use code with caution.
+Python
+
+requirements.txt:
+
+streamlit
+streamlit-webrt
+content_copy
+Use code with caution.