app.py

import os
import faiss
import numpy as np
import streamlit as st
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores.faiss import FAISS
from langchain.document_loaders import PyPDFLoader
from sentence_transformers import SentenceTransformer
from groq import Groq

# Initialize Groq API Client
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
client = Groq(api_key=GROQ_API_KEY)

# Load sentence-transformer model (open-source embedding model)
embedding_model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")

# Function to process PDF and create embeddings
def process_pdf(pdf_file):
    loader = PyPDFLoader(pdf_file)
    documents = loader.load()

    # Chunk text
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=512, chunk_overlap=50)
    chunks = text_splitter.split_documents(documents)

    # Generate embeddings
    texts = [chunk.page_content for chunk in chunks]
    embeddings = embedding_model.encode(texts, convert_to_numpy=True)

    # Store embeddings in FAISS
    dimension = embeddings.shape[1]
    index = faiss.IndexFlatL2(dimension)
    index.add(embeddings)

    faiss.write_index(index, "faiss_index")

    return texts, index

# Function to search in FAISS
def search_faiss(query, texts, index, top_k=3):
    query_embedding = embedding_model.encode([query], convert_to_numpy=True)
    D, I = index.search(query_embedding, top_k)

    results = [texts[i] for i in I[0] if i != -1]
    return results

# Function to retrieve response from Groq API
def get_groq_response(query):
    chat_completion = client.chat.completions.create(
        messages=[{"role": "user", "content": query}],
        model="llama-3.3-70b-versatile"
    )
    return chat_completion.choices[0].message.content

# Streamlit UI
st.title("RAG-Based PDF Chatbot")

uploaded_file = st.file_uploader("Upload a PDF", type="pdf")
if uploaded_file:
    with open("uploaded_document.pdf", "wb") as f:
        f.write(uploaded_file.getbuffer())

    st.success("PDF uploaded successfully!")
    texts, index = process_pdf("uploaded_document.pdf")

    query = st.text_input("Ask a question about the document:")
    if query:
        docs = search_faiss(query, texts, index)
        context = " ".join(docs)

        final_query = f"Context: {context} \n\n Question: {query}"
        response = get_groq_response(final_query)

        st.write("Response:", response)