Update app.py

app.py

@@ -3,119 +3,116 @@ import tempfile
 import gc
 import logging
 import streamlit as st
-from groq import Groq
+from groq import Groq, APIError
 from langchain_community.document_loaders import PyPDFLoader
 from langchain_text_splitters import RecursiveCharacterTextSplitter
 from langchain_community.embeddings import HuggingFaceEmbeddings
 from langchain_community.vectorstores import Chroma
 import torch
 
-# ---------------- CONFIG ----------------
+# ---------------- CONFIGURATION ----------------
 logging.basicConfig(level=logging.INFO)
 
+# Load API key from Hugging Face secrets
 GROQ_API_KEY = st.secrets.get("GROQ_API_KEY", os.environ.get("GROQ_API_KEY"))
 GROQ_MODEL = "llama-3.1-8b-instant"
 
+# Initialize Groq client
 client = None
 if GROQ_API_KEY:
     try:
         client = Groq(api_key=GROQ_API_KEY)
+        logging.info("✅ Groq client initialized successfully.")
     except Exception as e:
-        st.error(e)
+        st.error(f"❌ Failed to initialize Groq client: {e}")
+        client = None
+else:
+    st.warning("⚠️ GROQ_API_KEY not found. Please add it to Hugging Face secrets.")
 
-# ---------------- PAGE CONFIG ----------------
-st.set_page_config(
-    page_title="PDF Assistant",
-    page_icon="📘",
-    layout="wide"
-)
+# ---------------- STREAMLIT UI SETUP ----------------
+st.set_page_config(page_title="PDF Assistant", page_icon="📘", layout="wide")
 
-# ---------------- FORCE DARK UI (THEME INDEPENDENT) ----------------
+# ---------------- CSS ----------------
 st.markdown("""
 <style>
-:root {
-    --primary-color: #1e3a8a;
-    --background-color: #0e1117;
-    --secondary-background-color: #161b22;
-    --text-color: #f0f2f6;
+/* 1. GLOBAL RESET & SCROLL LOCK */
+html, body {
+    overflow: hidden;
+    height: 100%;
+    margin: 0;
 }
-
-html, body, [data-testid="stAppViewContainer"], .main {
-    background-color: #0e1117 !important;
-    color: #f0f2f6 !important;
+/* 2. HIDE DEFAULT STREAMLIT ELEMENTS */
+header[data-testid="stHeader"] {
+    display: none;
 }
-
-/* Hide Streamlit default UI */
-header[data-testid="stHeader"], footer {
+footer {
     display: none;
 }
-
-/* Sidebar */
+/* 3. SIDEBAR STYLING (INDEPENDENT LEFT PANEL SCROLL) */
 [data-testid="stSidebar"] {
-    background-color: #161b22 !important;
-    width: 20rem;
     position: fixed;
+    top: 0;
+    left: 0;
     height: 100vh;
-    overflow-y: auto;
+    width: 20rem;
+    overflow-y: auto !important;
+    z-index: 99999;
 }
-
-/* Fixed header */
+[data-testid="stSidebar"]::-webkit-scrollbar {
+    width: 6px;
+}
+[data-testid="stSidebar"]::-webkit-scrollbar-thumb {
+    background: #2d3748;
+    border-radius: 3px;
+}
+/* 4. FIXED HEADER STYLING */
 .fixed-header {
     position: fixed;
     top: 0;
     left: 0;
     width: 100%;
     height: 6rem;
-    background: #0e1117;
-    z-index: 9999;
+    background-color: #0e1117; 
+    z-index: 99998;
     display: flex;
     flex-direction: column;
-    align-items: center;
     justify-content: center;
-    border-bottom: 1px solid rgba(255,255,255,0.1);
-}
-
-.title-text {
-    font-size: 2.5rem;
-    font-weight: 800;
-}
-
-.creator-text {
-    font-size: 1rem;
-    color: #9ca3af;
-}
-
-.creator-text a {
-    color: #4da6ff;
-    text-decoration: none;
+    align-items: center;
+    border-bottom: 1px solid rgba(255, 255, 255, 0.1);
 }
-
-/* Main content spacing */
+/* 5. MAIN CONTENT SCROLLING (INDEPENDENT RIGHT PANEL SCROLL) */
 .main .block-container {
     margin-top: 6rem;
+    height: calc(100vh - 6rem);
+    overflow-y: auto;
+    padding-top: 1rem;
     padding-bottom: 5rem;
 }
-
-/* Inputs */
-input, textarea {
-    background-color: #1f2933 !important;
-    color: #f0f2f6 !important;
-    border: 1px solid #374151 !important;
+.main .block-container::-webkit-scrollbar {
+    width: 8px;
 }
-
-/* Buttons */
-button {
-    background-color: #1e3a8a !important;
-    color: white !important;
-    border-radius: 8px !important;
+.main .block-container::-webkit-scrollbar-thumb {
+    background: #2d3748;
+    border-radius: 4px;
+}
+/* 6. SIDEBAR BUTTON STYLING */
+[data-testid="stSidebar"] .stButton button {
+    width: 100%;
+    border-radius: 8px;
     font-weight: 600;
+    margin-bottom: 6px;
 }
-
-button:hover {
-    background-color: #2563eb !important;
+/* 7. HIDE UPLOADED FILE LIST & NAME */
+[data-testid='stFileUploaderFile'] {
+    display: none;
 }
-
-/* Chat bubbles */
+section[data-testid="stFileUploader"] ul {
+    display: none;
+}
+section[data-testid="stFileUploader"] small {
+    display: none;
+}
+/* 8. CHAT BUBBLES */
 .chat-user {
     background: #2d3748;
     padding: 12px;
@@ -123,48 +120,62 @@ button:hover {
     margin: 6px 0 6px auto;
     max-width: 85%;
     text-align: right;
+    color: #f0f2f6;
 }
-
 .chat-bot {
     background: #1e3a8a;
     padding: 12px;
     border-radius: 10px 10px 10px 2px;
     margin: 6px auto 6px 0;
     max-width: 85%;
+    text-align: left;
+    color: #ffffff;
 }
-
-/* Hide file list */
-[data-testid="stFileUploaderFile"],
-section[data-testid="stFileUploader"] ul,
-section[data-testid="stFileUploader"] small {
-    display: none;
+/* Sources CSS removed/hidden as it is no longer used */
+.sources {
+    display: none; 
 }
-
-/* Scrollbar */
-*::-webkit-scrollbar {
-    width: 8px;
+/* 9. TITLE TEXT */
+.title-text {
+    font-size: 2.5rem;
+    font-weight: 800;
+    margin: 0;
+    line-height: 1.2;
 }
-*::-webkit-scrollbar-thumb {
-    background: #2d3748;
-    border-radius: 4px;
+.creator-text {
+    font-size: 1rem;
+    font-weight: 500;
+    color: #cccccc;
+}
+.creator-text a {
+    color: #4da6ff;
+    text-decoration: none;
+}
+/* 10. INPUT FORM STYLING */
+[data-testid="stForm"] {
+    border: none;
+    padding: 0;
 }
 </style>
 """, unsafe_allow_html=True)
 
-# ---------------- HEADER ----------------
+# ---------------- FIXED HEADER ----------------
 st.markdown("""
 <div class="fixed-header">
     <div class="title-text">📘 PDF Assistant</div>
     <div class="creator-text">
-        by <a href="https://www.linkedin.com/in/abhishek-iitr/" target="_blank">
-        Abhishek Saxena</a>
+        by <a href="https://www.linkedin.com/in/abhishek-iitr/" target="_blank">Abhishek Saxena</a>
     </div>
 </div>
 """, unsafe_allow_html=True)
 
+
 # ---------------- SESSION STATE ----------------
-if "messages" not in st.session_state:
-    st.session_state.messages = []
+if "chat" not in st.session_state:
+    st.session_state.chat = []
+
+if "vectorstore" not in st.session_state:
+    st.session_state.vectorstore = None
 
 if "retriever" not in st.session_state:
     st.session_state.retriever = None
@@ -176,10 +187,11 @@ if "uploader_key" not in st.session_state:
     st.session_state.uploader_key = 0
 
 # ---------------- FUNCTIONS ----------------
-def clear_chat():
-    st.session_state.messages = []
+def clear_chat_history():
+    st.session_state.chat = []
 
 def clear_memory():
+    st.session_state.vectorstore = None
     st.session_state.retriever = None
     st.session_state.uploaded_file_name = None
     st.session_state.uploader_key += 1
@@ -187,102 +199,132 @@ def clear_memory():
     if torch.cuda.is_available():
         torch.cuda.empty_cache()
 
-def process_pdf(uploaded):
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
-        tmp.write(uploaded.getvalue())
-        path = tmp.name
-
-    loader = PyPDFLoader(path)
-    docs = loader.load()
-
-    splitter = RecursiveCharacterTextSplitter(chunk_size=800, chunk_overlap=60)
-    chunks = splitter.split_documents(docs)
-
-    embeddings = HuggingFaceEmbeddings(
-        model_name="sentence-transformers/all-MiniLM-L6-v2",
-        model_kwargs={"device": "cpu"},
-        encode_kwargs={"normalize_embeddings": True}
-    )
-
-    vectorstore = Chroma.from_documents(chunks, embeddings)
-    st.session_state.retriever = vectorstore.as_retriever(search_kwargs={"k": 5})
-    os.unlink(path)
-
-def ask(question):
-    docs = st.session_state.retriever.invoke(question)
-    context = "\n\n".join(d.page_content for d in docs)
-
-    prompt = f"""
-You are a strict RAG assistant.
-Use ONLY the PDF content.
-If the answer is not found, say:
-"I cannot find this in the PDF."
-
-CONTEXT:
-{context}
-
-QUESTION:
-{question}
+def process_pdf(uploaded_file):
+    try:
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
+            tmp.write(uploaded_file.getvalue())
+            path = tmp.name
+        
+        loader = PyPDFLoader(path)
+        docs = loader.load()
+        
+        splitter = RecursiveCharacterTextSplitter(chunk_size=800, chunk_overlap=60)
+        chunks = splitter.split_documents(docs)
+        
+        embeddings = HuggingFaceEmbeddings(
+            model_name="sentence-transformers/all-MiniLM-L6-v2",
+            model_kwargs={"device": "cpu"},
+            encode_kwargs={"normalize_embeddings": True}
+        )
+        
+        vectorstore = Chroma.from_documents(chunks, embeddings)
+        retriever = vectorstore.as_retriever(search_kwargs={"k": 5})
+        
+        st.session_state.vectorstore = vectorstore
+        st.session_state.retriever = retriever
+        
+        if os.path.exists(path):
+            os.unlink(path)
+        
+        return len(chunks)
+    except Exception as e:
+        st.error(f"Error processing PDF: {str(e)}")
+        return None
+
+def ask_question(question):
+    if not client:
+        return None, 0, "Groq client is not initialized."
+    if not st.session_state.retriever:
+        return None, 0, "Upload PDF first."
+    
+    try:
+        docs = st.session_state.retriever.invoke(question)
+        context = "\n\n".join(d.page_content for d in docs)
+        
+        prompt = f"""
+You are a strict RAG Q&A assistant.
+Summarize the context based on user question and return best answer. If the answer is not found, reply: "I cannot find this in the PDF."
+ CONTEXT = {context}
+QUESTION = {question}
+Answer on your behalf, don't say based on the context...
 """
-
-    res = client.chat.completions.create(
-        model=GROQ_MODEL,
-        messages=[{"role": "user", "content": prompt}],
-        temperature=0.1
-    )
-    return res.choices[0].message.content.strip()
+        
+        response = client.chat.completions.create(
+            model=GROQ_MODEL,
+            messages=[
+                {"role": "system", "content": "Use only the PDF content."},
+                {"role": "user", "content": prompt}
+            ],
+            temperature=0.1
+        )
+        return response.choices[0].message.content.strip(), len(docs), None
+    except Exception as e:
+        return None, 0, f"Error: {str(e)}"
 
 # ---------------- SIDEBAR ----------------
 with st.sidebar:
+    st.write("") 
     if st.button("🗑️ Clear Chat History", use_container_width=True):
-        clear_chat()
-    if st.button("🔥 Clear PDF Memory", use_container_width=True):
-        clear_memory()
-        st.success("Memory cleared")
+        clear_chat_history()
+    if st.button("🔥 Clear PDF Memory", on_click=clear_memory, use_container_width=True):
+         st.success("Memory Cleared!")
 
     st.markdown("---")
-
+    
+    upload_label = "✅ PDF Uploaded!" if st.session_state.uploaded_file_name else "Upload PDF"
     uploaded = st.file_uploader(
-        "Upload PDF",
-        type=["pdf"],
-        key=st.session_state.uploader_key,
-        label_visibility="collapsed"
+        upload_label, type=["pdf"], key=st.session_state.uploader_key, label_visibility="collapsed"
     )
 
-    if uploaded and uploaded.name != st.session_state.uploaded_file_name:
-        with st.spinner("Processing PDF..."):
-            process_pdf(uploaded)
-            st.session_state.uploaded_file_name = uploaded.name
-            st.session_state.messages = []
-            st.success("PDF ready")
-
-    if not uploaded:
-        st.warning("⬆️ Upload a PDF to start chatting")
+    if uploaded:
+        if uploaded.name != st.session_state.uploaded_file_name:
+            st.session_state.uploaded_file_name = None
+            st.session_state.chat = []
+            with st.spinner(f"Processing '{uploaded.name}'..."):
+                chunks = process_pdf(uploaded)
+                if chunks:
+                    st.session_state.uploaded_file_name = uploaded.name
+                    st.success("✅ PDF Processed!") 
+                else:
+                    st.error("❌ Failed.")
+        else:
+            st.success(f"✅ **Active:** `{uploaded.name}`")
+    else:
+        st.warning("⬆️ Upload a PDF to start chatting!")
 
-# ---------------- INPUT ----------------
-disabled = not st.session_state.uploaded_file_name or not client
+# ---------------- INPUT AREA ----------------
+disabled_input = st.session_state.uploaded_file_name is None or client is None
 
-with st.form(key="chat_form", clear_on_submit=True):
-    col1, col2 = st.columns([0.85, 0.15])
-    with col1:
-        question = st.text_input(
-            "",
+# Input Form
+with st.form(key='chat_form', clear_on_submit=True):
+    col_input, col_btn = st.columns([0.85, 0.15], gap="small")
+    with col_input:
+        user_question = st.text_input(
+            "Ask a question", 
             placeholder="Ask a question about the loaded PDF...",
-            disabled=disabled
+            label_visibility="collapsed",
+            disabled=disabled_input
         )
-    with col2:
-        send = st.form_submit_button("➤", disabled=disabled)
+    with col_btn:
+        submit_btn = st.form_submit_button("➤", disabled=disabled_input, use_container_width=True)
 
-if send and question:
-    st.session_state.messages.append(("user", question))
+if submit_btn and user_question:
+    st.session_state.chat.append(("user", user_question))
     with st.spinner("Thinking..."):
-        answer = ask(question)
-        st.session_state.messages.append(("bot", answer))
+        answer, sources, error = ask_question(user_question)
+        if answer:
+            # CHANGED: Removed the 'Context Chunks Used' HTML from the message
+            bot_msg = answer
+            st.session_state.chat.append(("bot", bot_msg))
+        else:
+            st.session_state.chat.append(("bot", f"🔴 **Error:** {error}"))
     st.rerun()
 
-# ---------------- CHAT HISTORY ----------------
-for role, msg in reversed(st.session_state.messages):
-    if role == "user":
-        st.markdown(f"<div class='chat-user'>{msg}</div>", unsafe_allow_html=True)
-    else:
-        st.markdown(f"<div class='chat-bot'>{msg}</div>", unsafe_allow_html=True)
+# ---------------- CHAT HISTORY (REVERSED) ----------------
+if st.session_state.chat:
+    st.markdown("---")
+    for role, msg in reversed(st.session_state.chat):
+        if role == "user":
+            st.markdown(f"<div class='chat-user'>{msg}</div>", unsafe_allow_html=True)
+        else:
+            st.markdown(f"<div class='chat-bot'>{msg}</div>", unsafe_allow_html=True)