Update app.py

app.py

@@ -1,6 +1,3 @@
-%%writefile app.py
-
-print("--- Python script starting ---")
 import streamlit as st
 import os
 os.environ['TOKENIZERS_PARALLELISM'] = 'false'
@@ -11,7 +8,6 @@ if not os.path.exists('/app/huggingface_cache'):
     os.makedirs('/app/huggingface_cache', exist_ok=True)
 
 import langchain
-langchain.debug = False # Keep this off for production speed
 
 from dotenv import load_dotenv
 from pinecone import Pinecone
@@ -25,7 +21,6 @@ from langchain_core.output_parsers import StrOutputParser
 from langchain.retrievers import ContextualCompressionRetriever
 from langchain.retrievers.document_compressors import CohereRerank
 
-print("--- All imports successful ---")
 
 try:
     print("Step 1: Loading environment variables...")
@@ -37,9 +32,8 @@ try:
     print("Step 1: SUCCESS")
 
     st.set_page_config(page_title="Advanced RAG Chatbot", page_icon="🚀", layout="wide")
-    # Remove the st.title here, we can let the chat UI be the main focus or add it inside the main area
+    
 
-    # --- Custom CSS for Chat Bubbles (FROM YOUR OFFLINE APP) ---
     st.markdown("""
         <style>
             .chat-container {
@@ -96,13 +90,13 @@ try:
 
     @st.cache_resource
     def initialize_services():
-        # (Same as your previous working initialize_services function)
+        
         print("Step 2: Entering initialize_services function...")
         if not all([PINECONE_API_KEY, GROQ_API_KEY, COHERE_API_KEY]):
             raise ValueError("An API key is missing!")
         embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
         pinecone = Pinecone(api_key=PINECONE_API_KEY)
-        host = "https://rag-chatbot-sg8t88c.svc.aped-4627-b74a.pinecone.io" # Your host
+        host = "https://rag-chatbot-sg8t88c.svc.aped-4627-b74a.pinecone.io" 
         index = pinecone.Index(host=host)
         vectorstore = PineconeVectorStore(index=index, embedding=embeddings)
         base_retriever = vectorstore.as_retriever(search_kwargs={'k': 10})
@@ -116,7 +110,7 @@ try:
     retriever, llm = initialize_services()
     print("Step 3: SUCCESS, services are loaded.")
 
-    # --- RAG CHAIN (modified for streaming if possible, and simpler output) ---
+    # --- RAG CHAIN 
     print("Step 4: Defining RAG chain...")
     system_prompt = """You are a helpful AI assistant that answers questions based ONLY on the provided context.
     Your answer should be concise and directly address the question.
@@ -149,14 +143,13 @@ try:
     )
     print("Step 4: SUCCESS")
 
-    # --- Streamlit Chat UI (ADAPTED FROM YOUR OFFLINE APP) ---
-    st.title("💬 Document Chatbot Interface") # You can add a title here
+    # --- Streamlit Chat UI
+    st.title("💬 Document Chatbot Interface") 
 
     if "messages" not in st.session_state:
         st.session_state.messages = [{"role": "assistant", "content": "Hello! I'm ready to answer questions about your documents.", "sources": []}]
 
-    # Display chat messages using your custom HTML
-    # Wrap messages in a container for better layout control if needed
+   
     st.markdown('<div class="chat-container">', unsafe_allow_html=True)
     for message in st.session_state.messages:
         if message["role"] == "user":
@@ -185,7 +178,7 @@ try:
                 try:
                     print(f"--- UI DEBUG: Invoking RAG chain with query: {user_query} ---")
                     
-                    # For streaming with StrOutputParser, we iterate over chunks
+                   
                     assistant_response_content = ""
                     for chunk in rag_chain.stream(user_query):
                         assistant_response_content += chunk
@@ -194,7 +187,7 @@ try:
                     message_placeholder.markdown(f'<div class="chat-bubble bot-bubble">{assistant_response_content}</div>', unsafe_allow_html=True) # Final response
                     print(f"--- UI DEBUG: Full LLM Answer: {assistant_response_content} ---")
 
-                    # Retrieve sources separately for display (as before)
+                    
                     retrieved_docs_for_display = retriever.invoke(user_query)
                     sources_info_for_display = []
                     if retrieved_docs_for_display:
@@ -205,17 +198,14 @@ try:
                                 "content_snippet": doc.page_content
                             })
                     
-                    # Add assistant's full response and sources to session state
+                   
                     st.session_state.messages.append({
                         "role": "assistant", 
                         "content": assistant_response_content,
-                        "sources": sources_info_for_display # Store source info with the message
+                        "sources": sources_info_for_display 
                     })
 
-                    # Display sources in an expander (this will appear below the last message outside the chat_message context)
-                    # This part needs to be rethought slightly for a clean UI if sources are tied to each message
-                    # For now, let's keep the separate expander logic for the last response.
-                    # The main display loop will handle showing sources for historical messages.
+                    
                     if sources_info_for_display:
                         with st.expander("Sources for the latest answer"):
                             for i, doc_info in enumerate(sources_info_for_display):
@@ -223,8 +213,7 @@ try:
                                 st.markdown(f"> {doc_info['content_snippet'][:300]}...")
                                 st.markdown("---")
                     
-                    # A slight hack to force rerun to update the main message display loop with the new message + sources
-                    # st.experimental_rerun() # Not always needed if placeholders work well
+                    
 
                 except Exception as e_invoke:
                     error_message = f"Error processing your query: {e_invoke}"
@@ -235,9 +224,3 @@ try:
                     st.session_state.messages.append({"role": "assistant", "content": f"Sorry, I encountered an error: {error_message}", "sources": []})
     
     print("--- app.py script finished a run ---")
-
-except Exception as e:
-    print(f"!!!!!!!!!! A FATAL ERROR OCCURRED DURING STARTUP !!!!!!!!!!")
-    import traceback
-    print(traceback.format_exc())
-    st.error(f"A fatal error occurred during startup. Please check the container logs. Error: {e}")