Update app.py

app.py

@@ -2,66 +2,100 @@ import streamlit as st
 import os
 from llama_cpp import Llama
 
+# 1. PAGE SETUP & MOBILE STYLING
 st.set_page_config(page_title="Nova AI", page_icon="✨", layout="wide")
 
+st.markdown("""
+    <style>
+    div.block-container{padding-top:2rem;}
+    .stChatFloatingInputContainer{padding-bottom:1rem;}
+    </style>
+    """, unsafe_allow_html=True)
+
+# 2. SESSION STATE (The App's Memory)
 if "auth" not in st.session_state:
     st.session_state.auth = False
 if "messages" not in st.session_state:
     st.session_state.messages = []
 
-# --- 1. LOGIN ---
+# 3. SECURE LOGIN
 if not st.session_state.auth:
     st.title("🔐 Nova Secure Entry")
     with st.form("login"):
+        # Access your password from Settings > Secrets
+        SECRET_PASSWORD = os.getenv("NOVA_PASSWORD", "admin")
         pwd = st.text_input("Access Key", type="password")
-        if st.form_submit_button("Unlock"):
-            if pwd == os.getenv("NOVA_PASSWORD", "admin"):
+        if st.form_submit_button("Unlock Nova"):
+            if pwd == SECRET_PASSWORD:
                 st.session_state.auth = True
                 st.rerun()
             else:
                 st.error("Invalid Key")
     st.stop()
 
-# --- 2. FAST MODEL SETTINGS ---
+# 4. FAST MODEL SETTINGS (Optimized for 16GB RAM)
 @st.cache_resource
 def load_nova():
     return Llama.from_pretrained(
         repo_id="HuggingFaceTB/SmolLM2-1.7B-Instruct-GGUF",
         filename="*q4_k_m.gguf",
-        n_ctx=512,      # Small context = fast CPU speed
-        n_threads=4,    # Uses all available CPU power
-        n_batch=512,
+        n_ctx=2048,      # Large memory for long scripts
+        n_threads=4,     # Max CPU power
+        n_batch=512,     # Efficient processing
         verbose=False
     )
 
+with st.sidebar:
+    st.title("✧ Nova Cloud-Alpha")
+    st.info("Status: Online | Secure")
+    if st.button("Clear Chat History"):
+        st.session_state.messages = []
+        st.rerun()
+
 llm = load_nova()
 
-# --- 3. UI ---
+# 5. CHAT INTERFACE
 st.title("✨ Nova Coding Assistant")
 
+# Display previous messages with custom avatars
 for message in st.session_state.messages:
-    with st.chat_message(message["role"]):
+    avatar_sym = "< >" if message["role"] == "user" else "✨"
+    with st.chat_message(message["role"], avatar=avatar_sym):
         st.markdown(message["content"])
 
+# 6. AI GENERATION (With Status Bar & Streaming)
 if prompt := st.chat_input("Ask Nova..."):
+    # Add your message to history
     st.session_state.messages.append({"role": "user", "content": prompt})
-    with st.chat_message("user"):
+    with st.chat_message("user", avatar="< >"):
         st.markdown(prompt)
 
-    with st.chat_message("assistant"):
-        # The System prompt tells her to STOP the "As an AI" speeches
-        full_prompt = f"<|im_start|>system\nYou are Nova, a witty, fast, and direct coding assistant. No yapping. No 'As an AI' disclaimers. Just answer the user directly.<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n"
-        
-        # This part makes the words "stream" in real-time
-        response_container = st.empty()
-        full_response = ""
-        
-        # We loop through the tokens to show them one by one
-        for chunk in llm(full_prompt, max_tokens=256, stop=["<|im_end|>"], stream=True):
-            token = chunk["choices"][0]["text"]
-            full_response += token
-            response_container.markdown(full_response + "▌")
+    with st.chat_message("assistant", avatar="✨"):
+        # The Status Bar
+        with st.status("✨ Nova is processing...", expanded=True) as status:
+            st.write("Initializing neural links...")
+            
+            system_instr = "You are Nova, a witty, fast, and direct coding assistant. No yapping. No 'As an AI' disclaimers. Answer concisely."
+            full_prompt = f"<|im_start|>system\n{system_instr}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n"
+            
+            response_container = st.empty()
+            full_response = ""
+            
+            st.write("Generating code...")
+            
+            # Streaming the response word-by-word
+            for chunk in llm(
+                full_prompt, 
+                max_tokens=1024, # High lung capacity for long code
+                stop=["<|im_end|>", "User:"], 
+                stream=True
+            ):
+                token = chunk["choices"][0]["text"]
+                full_response += token
+                response_container.markdown(full_response + "▌") # Cursor effect
+            
+            status.update(label="✅ Nova Ready", state="complete", expanded=False)
         
+        # Final cleanup and save
         response_container.markdown(full_response)
-        st.session_state.messages.append({"role": "assistant", "content": full_response})
-        
+        st.session_state.messages.append({"role": "assistant", "content": full_response})