app.py

import streamlit as st
from groq import Groq
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import FAISS
from langchain.text_splitter import CharacterTextSplitter
from langchain.document_loaders import PyPDFLoader
from langchain.chains import RetrievalQA
from langchain.chat_models import ChatOpenAI
import tempfile
import os

# Set API Key
GROQ_API_KEY = os.getenv("your_groq_api_key")

# Initialize Groq client
client = Groq(api_key=GROQ_API_KEY)

# Streamlit App Title
st.title("📄 RAG Chatbot with Using Deepseek")

# File Upload
uploaded_files = st.file_uploader("📂 Upload PDFs", type=["pdf"], accept_multiple_files=True)

if uploaded_files:
    with st.spinner("Processing PDFs... ⏳"):
        docs = []
        for file in uploaded_files:
            # Save uploaded file temporarily
            with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as temp_pdf:
                temp_pdf.write(file.read())
                temp_pdf_path = temp_pdf.name
            
            # Load PDF and extract text
            loader = PyPDFLoader(temp_pdf_path)
            docs.extend(loader.load())

            # Clean up temp file
            os.remove(temp_pdf_path)
        
        # Split text into smaller chunks
        text_splitter = CharacterTextSplitter(chunk_size=500, chunk_overlap=50)
        split_docs = text_splitter.split_documents(docs)
        
        # Create embeddings & FAISS vector store
        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
        vector_db = FAISS.from_documents(split_docs, embeddings)
        
        st.success("✅ PDFs processed and stored in the vector database!")

    # User Input
    query = st.text_input("🔍 Ask a question:")
    
    if query:
        with st.spinner("Retrieving answer... ⏳"):
            # Perform retrieval
            retriever = vector_db.as_retriever()
            relevant_docs = retriever.get_relevant_documents(query)
            
            # Get context from retrieved docs
            context = "\n\n".join([doc.page_content for doc in relevant_docs])
            
            # Make API call to Groq LLaMA-70B
            completion = client.chat.completions.create(
                model="deepseek-r1-distill-llama-70b",
                messages=[
                    {"role": "system", "content": "You are an AI assistant providing answers based on the given context."},
                    {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {query}"}
                ],
                temperature=0.6,
                max_completion_tokens=1024,
                top_p=0.95,
                stream=True,
                reasoning_format="raw"
            )

            # Stream response
            response_text = ""
            response_container = st.empty()
            for chunk in completion:
                response_text += chunk.choices[0].delta.content or ""
                response_container.markdown(response_text)

            st.success("✅ Answer generated!")