Update app.py

app.py

@@ -1,23 +1,71 @@
 # ---- Standard libs ----
 import os, tempfile
 from datetime import datetime
-import json
 
 # ---- Streamlit ----
 import streamlit as st
 
 # ---- Load environment variables (.env) ----
 from dotenv import load_dotenv
-load_dotenv()
+load_dotenv()                        # <-- makes GROQ_API_KEY available
 
 # ---- RAG + Groq stack ----
 from langchain_groq import ChatGroq
 from langchain.embeddings import HuggingFaceEmbeddings
 from langchain_community.vectorstores import Chroma
 from langchain.chains import RetrievalQA
-from langchain_community.document_loaders import PyPDFLoader
+from langchain_community.document_loaders import UnstructuredFileLoader
 from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain_chroma import Chroma
+
+
+# ------------- One‑time chain initialiser -------------
+@st.cache_resource(show_spinner=False)
+def init_rag_chain():
+    # 1️⃣ Embeddings
+    embedding = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
+
+    # 2️⃣ Persisted vector DB
+    vectordb = Chroma(
+        persist_directory="./rag_db",
+        embedding_function=embedding
+    )
+
+
+    
+    llm = ChatGroq(
+    groq_api_key=os.getenv("GROQ_API_KEY"),  # <--- THIS is critical
+    model="llama3-70b-8192",
+    temperature=0.2,
+    )
+
+
+    # 4️⃣ Retrieval‑augmented QA chain
+    return RetrievalQA.from_chain_type(
+        llm=llm,
+        retriever=vectordb.as_retriever(search_kwargs={"k": 4}),
+        return_source_documents=False
+    )
+
+qa_chain = init_rag_chain()
+
+# ------------- Optional: live‑ingest a PDF the user uploads -------------
+def ingest_pdf(file_bytes: bytes, file_name: str):
+    embedding = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
+    vectordb = Chroma(
+        persist_directory="./rag_db",
+        embedding_function=embedding
+    )
+
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
+        tmp.write(file_bytes)
+        tmp_path = tmp.name
+
+    docs = UnstructuredFileLoader(tmp_path).load()
+    splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
+    vectordb.add_documents(splitter.split_documents(docs))
+    vectordb.persist()
+
+
 # ------------- Configuration -------------
 st.set_page_config(
     page_title="📄 Multimodal Chat Assistant",
@@ -170,294 +218,32 @@ st.markdown("""
             margin-bottom: 2rem;
         }
     </style>
-""", unsafe_allow_html=True)
-
-# ------------- One‑time chain initializer -------------
-@st.cache_resource(show_spinner=False)
-def init_rag_chain(model_name="llama3-70b-8192", temperature=0.2):
-    embedding = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
-    vectordb = Chroma(
-        persist_directory="./rag_db",
-        embedding_function=embedding,
-        collection_metadata={"hnsw:space": "cosine"}
-    )
-    llm = ChatGroq(
-        groq_api_key=os.getenv("GROQ_API_KEY"),
-        model=model_name,
-        temperature=temperature,
-    )
-    return RetrievalQA.from_chain_type(
-        llm=llm,
-        retriever=vectordb.as_retriever(search_kwargs={"k": 4}),
-        return_source_documents=False
-    )
-
-# ------------- Document ingestion -------------
-  # updated import ✅
-
-def ingest_pdf(file_bytes: bytes, file_name: str):
-    embedding = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
+""", unsafe_allow_html=True)    
 
-    vectordb = Chroma(
-        persist_directory="./rag_db",
-        embedding_function=embedding,
-        collection_metadata={"hnsw:space": "cosine"}
-    )
-
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
-        tmp.write(file_bytes)
-        tmp_path = tmp.name
-
-    loader = PyPDFLoader(tmp_path)
-    docs = loader.load()
-
-    splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
-    chunks = splitter.split_documents(docs)
-
-    os.unlink(tmp_path)
-
-    if not chunks:
-        st.error("❌ No text chunks extracted from the PDF.")
-        return 0
-
-    # Remove empty chunks
-    non_empty_chunks = [chunk for chunk in chunks if chunk.page_content.strip()]
-    if not non_empty_chunks:
-        st.error("❌ All extracted chunks are empty. Check your PDF content.")
-        return 0
-
-    vectordb.add_documents(non_empty_chunks)
-    vectordb.persist()
-
-    if "documents" not in st.session_state:
-        st.session_state.documents = []
-
-    st.session_state.documents.append({
-        "name": file_name,
-        "chunks": len(non_empty_chunks),
-        "uploaded_at": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
-    })
-
-    return len(non_empty_chunks)
-
-
-# ------------- Initialize session state -------------
-if "chat_history" not in st.session_state:
-    st.session_state.chat_history = []
-if "documents" not in st.session_state:
-    st.session_state.documents = []
-if "total_questions" not in st.session_state:
-    st.session_state.total_questions = 0
-
-
-# ------------- Sidebar Configuration -------------
-with st.sidebar:
-    st.markdown("### ⚙️ Configuration")
-    
-    # Model selection
-    model_options = {
-        "Llama 3 70B": "llama3-70b-8192",
-        "Llama 3 8B": "llama3-8b-8192",
-        "Mixtral 8x7B": "mixtral-8x7b-32768",
-        "Gemma 7B": "gemma-7b-it"
-    }
-    
-    selected_model = st.selectbox(
-        "🤖 Select Model",
-        options=list(model_options.keys()),
-        index=0
-    )
-    
-    temperature = st.slider(
-        "🌡️ Temperature",
-        min_value=0.0,
-        max_value=1.0,
-        value=0.2,
-        step=0.1,
-        help="Higher values make output more creative"
-    )
-    
-    # Document statistics
-    st.markdown("### 📊 Statistics")
-    
-    col1, col2 = st.columns(2)
-    with col1:
-        st.markdown(f"""
-        <div class="stats-card">
-            <h4>📄 Documents</h4>
-            <h2>{len(st.session_state.documents)}</h2>
-        </div>
-        """, unsafe_allow_html=True)
-    
-    with col2:
-        st.markdown(f"""
-        <div class="stats-card">
-            <h4>💬 Questions</h4>
-            <h2>{st.session_state.total_questions}</h2>
-        </div>
-        """, unsafe_allow_html=True)
-    
-    # Document list
-    if st.session_state.documents:
-        st.markdown("### 📚 Uploaded Documents")
-        for doc in st.session_state.documents:
-            st.markdown(f"""
-            <div class="feature-card">
-                <strong>{doc['name']}</strong><br>
-                <small>📄 {doc['chunks']} chunks</small><br>
-                <small>🕒 {doc['uploaded_at']}</small>
-            </div>
-            """, unsafe_allow_html=True)
-    
-    # Clear chat button
-    if st.button("🗑️ Clear Chat History", use_container_width=True):
-        st.session_state.chat_history = []
-        st.rerun()
-
-# ------------- Main Interface -------------
-st.markdown('<h1 class="premium-title">🤖 Multimodal Chat Assistant</h1>', unsafe_allow_html=True)
-st.markdown('<p class="subtitle">Upload your documents and chat with AI-powered intelligence</p>', unsafe_allow_html=True)
-
-# Initialize QA chain with selected model
-qa_chain = init_rag_chain(model_options[selected_model], temperature)
 
 # Upload section
-col1, col2, col3 = st.columns([1, 2, 1])
-with col2:
-    st.markdown('<div class="upload-area">', unsafe_allow_html=True)
-    uploaded_file = st.file_uploader(
-        "📤 Upload a PDF file",
-        type=["pdf"],
-        help="Upload a PDF document to start chatting with it"
-    )
-    st.markdown('</div>', unsafe_allow_html=True)
-
+uploaded_file = st.file_uploader("📤 Upload a PDF file", type=["pdf"])
 if uploaded_file is not None:
-    with st.spinner("📚 Processing document... This may take a moment."):
-        progress_bar = st.progress(0)
-        for i in range(100):
-            progress_bar.progress(i + 1)
-        
-        chunks_count = ingest_pdf(uploaded_file.read(), uploaded_file.name)
-        progress_bar.empty()
-        
-        st.success(f"✅ Document processed successfully! Created {chunks_count} chunks.")
-        st.balloons()
+    with st.spinner("📚 Ingesting document..."):
+        ingest_pdf(uploaded_file.read(), uploaded_file.name)
+    st.success("✅ Document ingested. Ask away!")
 
-# Chat interface
-st.markdown('<div class="chat-container">', unsafe_allow_html=True)
+# Chat UI
+if "chat_history" not in st.session_state:
+    st.session_state.chat_history = []
 
-# Suggested questions
-if not st.session_state.chat_history and st.session_state.documents:
-    st.markdown("### 💡 Suggested Questions")
-    suggestions = [
-        "📋 What is the main topic of this document?",
-        "🔍 Can you summarize the key points?",
-        "📊 What are the important statistics mentioned?",
-        "🎯 What conclusions does the document draw?"
-    ]
-    
-    cols = st.columns(2)
-    for i, suggestion in enumerate(suggestions):
-        with cols[i % 2]:
-            if st.button(suggestion, key=f"suggestion_{i}", use_container_width=True):
-                st.session_state.chat_history.append({
-                    "role": "user", 
-                    "content": suggestion.split(" ", 1)[1],
-                    "timestamp": datetime.now().strftime("%H:%M")
-                })
-                st.session_state.total_questions += 1
-                st.rerun()
-
-# Chat input
-prompt = st.text_input(
-    "💬 Ask a question about your document",
-    placeholder="Type your question here...",
-    key="chat_input"
-)
+prompt = st.text_input("💬 Ask a question about the document")
 
 if prompt:
-    # Add user message
-    st.session_state.chat_history.append({
-        "role": "user", 
-        "content": prompt,
-        "timestamp": datetime.now().strftime("%H:%M")
-    })
-    st.session_state.total_questions += 1
-    
-    # Generate response
+    st.session_state.chat_history.append({"role": "user", "content": prompt})
+
     with st.spinner("🤔 Thinking..."):
-        try:
-            response = qa_chain.run(prompt)
-            st.session_state.chat_history.append({
-                "role": "assistant", 
-                "content": response,
-                "timestamp": datetime.now().strftime("%H:%M")
-            })
-        except Exception as e:
-            st.error(f"❌ Error: {str(e)}")
-    
-    st.rerun()
+        response = qa_chain.run(prompt)
+        st.session_state.chat_history.append({"role": "assistant", "content": response})
 
 # Display chat history
-if st.session_state.chat_history:
-    st.markdown("### 💬 Conversation")
-    
-    for i, msg in enumerate(st.session_state.chat_history):
-        if msg["role"] == "user":
-            st.markdown(f"""
-            <div style="display: flex; justify-content: flex-end; margin: 1rem 0;">
-                <div class="message-user">
-                    <strong>🧑 You</strong> <small style="opacity: 0.7;">({msg.get('timestamp', '')})</small><br>
-                    {msg['content']}
-                </div>
-            </div>
-            """, unsafe_allow_html=True)
-        else:
-            st.markdown(f"""
-            <div style="display: flex; justify-content: flex-start; margin: 1rem 0;">
-                <div class="message-bot">
-                    <strong>🤖 Assistant</strong> <small style="opacity: 0.7;">({msg.get('timestamp', '')})</small><br>
-                    {msg['content']}
-                </div>
-            </div>
-            """, unsafe_allow_html=True)
-            
-            # Add action buttons for bot responses
-            col1, col2, col3, col4 = st.columns([1, 1, 1, 5])
-            with col1:
-                if st.button("👍", key=f"like_{i}", help="Like this response"):
-                    st.toast("👍 Response liked!")
-            with col2:
-                if st.button("👎", key=f"dislike_{i}", help="Dislike this response"):
-                    st.toast("👎 Feedback noted!")
-            with col3:
-                if st.button("📋", key=f"copy_{i}", help="Copy response"):
-                    st.toast("📋 Response copied!")
-
-else:
-    if not st.session_state.documents:
-        st.markdown("""
-        <div class="feature-card" style="text-align: center;">
-            <h3>🚀 Get Started</h3>
-            <p>Upload a PDF document above to begin chatting with your AI assistant!</p>
-            
-            <h4>✨ Features:</h4>
-            <ul style="text-align: left; display: inline-block;">
-                <li>🤖 Multiple AI models to choose from</li>
-                <li>📄 Advanced document processing</li>
-                <li>💬 Interactive chat interface</li>
-                <li>📊 Real-time statistics</li>
-                <li>🎨 Premium UI/UX design</li>
-            </ul>
-        </div>
-        """, unsafe_allow_html=True)
-
-st.markdown('</div>', unsafe_allow_html=True)
-
-# Footer
-st.markdown("""
-<div style="text-align: center; margin-top: 3rem; padding: 2rem; color: rgba(255,255,255,0.6);">
-    <p>🤖 Powered by Groq AI • Built with ❤️ using Streamlit</p>
-</div>
-""", unsafe_allow_html=True)
+for msg in st.session_state.chat_history:
+    if msg["role"] == "user":
+        st.markdown(f"<div class='message-bubble'><b>🧑 You:</b> {msg['content']}</div>", unsafe_allow_html=True)
+    else:
+        st.markdown(f"<div class='message-bubble bot'><b>🤖 Bot:</b> {msg['content']}</div>", unsafe_allow_html=True)