Update app.py

app.py

@@ -6,16 +6,14 @@ import time
 import mimetypes
 from pathlib import Path
 
-# Sidebar API key input
-with st.sidebar:
-    user_api_key = st.text_input("Enter your Gemini API key (optional):", type="password")
+# Step 1: Enter API Key
+st.sidebar.header("Step 1: Enter Gemini API Key")
+user_api_key = st.sidebar.text_input("Enter your Gemini API key (optional):", type="password")
 
 # Gemini API key configuration
 GEMINI_API_KEY = user_api_key if user_api_key else "AIzaSyDCMPwXHagWqYTQB3HL7FceHEmKUv3v4wc"
 genai.configure(api_key=GEMINI_API_KEY)
 
-#st.set_page_config(page_title="Video-Based Retrieval-Augmented Generation", page_icon="🎬", layout="wide")
-
 class VideoProcessor:
     def __init__(self):
         self.model = genai.GenerativeModel("gemini-2.0-flash")
@@ -40,46 +38,47 @@ for key in ["video_processor", "video_file", "video_name", "messages"]:
     if key not in st.session_state:
         st.session_state[key] = None if key != "messages" else []
 
-with st.sidebar:
-    st.header("📹 Upload Video")
-    uploaded_file = st.file_uploader("Choose video", type=['mp4', 'mov', 'avi', 'mkv', 'webm'])
+st.title("🎬 Video Retrieval-Augmented Generation - Gemini 2.0")
 
-    if uploaded_file:
-        if mimetypes.guess_type(uploaded_file.name)[0].startswith("video"):
-            file_size = len(uploaded_file.getvalue()) / (1024**2)
-            st.info(f"Size: {file_size:.2f} MB")
+# Step 2: Upload Video
+st.header("Step 2: Upload Video")
+uploaded_file = st.file_uploader("Choose a video file", type=['mp4', 'mov', 'avi', 'mkv', 'webm'])
 
-            if st.session_state.video_name != uploaded_file.name:
-                st.session_state.video_processor = VideoProcessor()
-                with tempfile.NamedTemporaryFile(delete=False, suffix=Path(uploaded_file.name).suffix) as tmp:
-                    tmp.write(uploaded_file.getvalue())
-                    tmp_path = tmp.name
+if uploaded_file:
+    if mimetypes.guess_type(uploaded_file.name)[0].startswith("video"):
+        file_size = len(uploaded_file.getvalue()) / (1024**2)
+        st.info(f"Size: {file_size:.2f} MB")
 
-                with st.spinner("Uploading and processing..."):
-                    video_file = st.session_state.video_processor.upload_video(tmp_path, uploaded_file.name)
-                    processed_file = st.session_state.video_processor.wait_for_processing(video_file)
-                    st.session_state.video_file = processed_file
-                    st.session_state.video_name = uploaded_file.name
-                    st.session_state.messages.clear()
-                    st.success("✅ Video processed")
+        if st.session_state.video_name != uploaded_file.name:
+            st.session_state.video_processor = VideoProcessor()
+            with tempfile.NamedTemporaryFile(delete=False, suffix=Path(uploaded_file.name).suffix) as tmp:
+                tmp.write(uploaded_file.getvalue())
+                tmp_path = tmp.name
 
-                os.unlink(tmp_path)
+            with st.spinner("Uploading and processing..."):
+                video_file = st.session_state.video_processor.upload_video(tmp_path, uploaded_file.name)
+                processed_file = st.session_state.video_processor.wait_for_processing(video_file)
+                st.session_state.video_file = processed_file
+                st.session_state.video_name = uploaded_file.name
+                st.session_state.messages.clear()
+                st.success("✅ Video processed")
 
-            st.video(uploaded_file.getvalue())
-        else:
-            st.error("Not a valid video file")
+            os.unlink(tmp_path)
 
-    if st.button("Reset Chat"):
-        st.session_state.messages.clear()
+        st.video(uploaded_file.getvalue())
+    else:
+        st.error("Not a valid video file")
 
-    if st.button("Reset All"):
-        st.session_state.clear()
+if st.button("Reset Chat"):
+    st.session_state.messages.clear()
 
-st.title("Video Retrieval-Augmented Generation - Gemini 2.0")
+if st.button("Reset All"):
+    st.session_state.clear()
+
+# Step 3: Chatting about the Video
+if st.session_state.video_file:
+    st.header("Step 3: Chat with Video")
 
-if not st.session_state.video_file:
-    st.info("👈 Upload a video to start chatting")
-else:
     for msg in st.session_state.messages:
         with st.chat_message(msg["role"]):
             st.markdown(msg["content"])
@@ -98,3 +97,5 @@ else:
 
             placeholder.markdown(response)
             st.session_state.messages.append({"role": "assistant", "content": response})
+else:
+    st.info("👈 Please upload a video to start chatting.")