app.py

import os
import streamlit as st
from groq import Groq
from langchain.chains import RetrievalQA
from langchain.vectorstores import FAISS
from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from io import BytesIO

# Set up Groq API key
GROQ_API_KEY = "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"

# Define a custom embedding class for Groq
class GroqEmbedding:
    def __init__(self, model="groq-embedding-model"):
        self.model = model
        self.client = Groq(api_key=GROQ_API_KEY)

    def embed_documents(self, texts):
        # Use Groq's API to generate embeddings for documents
        embeddings = self.client.embed_documents(texts, model=self.model)
        return embeddings
    
    def embed_query(self, query):
        # Use Groq's API to generate embedding for a query
        return self.client.embed_query(query, model=self.model)

# Streamlit App UI
st.title("PDF Question-Answering with Groq Embeddings")

uploaded_file = st.file_uploader("Upload a PDF", type="pdf")

# Process the uploaded PDF
if uploaded_file is not None:
    # Convert the uploaded file to a BytesIO object to read it in-memory
    pdf_file = BytesIO(uploaded_file.read())
    
    # Load the PDF file with PyPDFLoader
    loader = PyPDFLoader(pdf_file)
    documents = loader.load()
    
    # Split documents into smaller chunks for better processing
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
    split_docs = text_splitter.split_documents(documents)

    # Create embeddings using Groq
    embeddings = GroqEmbedding(model="groq-embedding-model")  # Use your preferred Groq model
    
    # Create a FAISS vector store with the embeddings
    vector_db = FAISS.from_documents(split_docs, embeddings)
    
    # Initialize the retrieval-based QA system
    qa = RetrievalQA.from_chain_type(llm=None, chain_type="stuff", vectorstore=vector_db)
    
    # User input for querying the PDF content
    query = st.text_input("Ask a question about the PDF:")

    if query:
        result = qa.run(query)
        st.write("Answer:", result)