app.py

import gradio as gr
import os
from langchain_community.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.vectorstores import FAISS
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from langchain_community.embeddings import HuggingFaceEmbeddings
from groq import Groq  # Using Groq's direct client instead of LangChain wrapper
from dotenv import load_dotenv

# Initialize Groq client with your direct API key
GROQ_API_KEY = "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"  # Paste your actual Groq API key here
groq_client = Groq(api_key=GROQ_API_KEY)

def process_pdf(file):
    try:
        # Save uploaded file
        filepath = file.name

        # Load PDF
        loader = PyPDFLoader(filepath)
        documents = loader.load()

        # Split into chunks
        text_splitter = RecursiveCharacterTextSplitter(
            chunk_size=1000,
            chunk_overlap=200,
            separators=["\n\n", "\n", " ", ""]
        )
        docs = text_splitter.split_documents(documents)

        # Create vector store with HuggingFace embeddings
        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
        vectorstore = FAISS.from_documents(docs, embeddings)

        # Save vectorstore for later use
        vectorstore.save_local("faiss_index")

        # Return the first chunk for preview and document info
        doc_info = f"Processed {len(docs)} chunks from {len(documents)} pages."
        return docs[0].page_content, doc_info
    except Exception as e:
        return f"Error: {str(e)}", ""

def answer_question(question):
    try:
        # Load vector store
        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
        vectorstore = FAISS.load_local("faiss_index", embeddings, allow_dangerous_deserialization=True)

        # Create retriever
        retriever = vectorstore.as_retriever(search_kwargs={"k": 3})

        # Get relevant documents
        docs = retriever.invoke(question)
        context = "\n\n".join([doc.page_content for doc in docs])

        # Create prompt template for LLaMA 3
        prompt = f"""<|begin_of_text|><|start_header_id|>system<|end_header_id|>
        You are a helpful AI assistant that answers questions based on the provided context.
        Use only the information from the context to answer the question. If you don't know the answer, say you don't know.
        Be concise and accurate in your responses.<|eot_id|>
        <|start_header_id|>user<|end_header_id|>
        Context: {context}
        Question: {question}<|eot_id|>
        <|start_header_id|>assistant<|end_header_id|>"""

        # Get response from Groq
        response = groq_client.chat.completions.create(
            messages=[{"role": "user", "content": prompt}],
            model="llama3-70b-8192",
            temperature=0,
        )

        return response.choices[0].message.content
    except Exception as e:
        return f"Error: {str(e)}"

# Gradio UI
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown("## 📄 PDF RAG System with Groq & LLaMA 3", elem_id="title")
    gr.Markdown("Upload a PDF file, process it, and ask questions using LLaMA 3. ✨")

    with gr.Tab("📂 Upload & Process"):
        with gr.Row():
            with gr.Column():
                pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
                process_button = gr.Button("🚀 Process PDF", variant="primary")
                doc_info = gr.Textbox(label="Document Info")
            output_text = gr.Textbox(
                label="First Chunk Preview",
                lines=12,
                placeholder="Processed PDF content will appear here...",
                show_copy_button=True
            )
        process_button.click(
            fn=process_pdf,
            inputs=pdf_input,
            outputs=[output_text, doc_info]
        )

    with gr.Tab("❓ Ask Questions"):
        with gr.Row():
            with gr.Column():
                question_input = gr.Textbox(
                    label="Your Question",
                    placeholder="Ask something about the PDF content..."
                )
                ask_button = gr.Button("🔍 Get Answer", variant="primary")
            answer_output = gr.Textbox(
                label="Answer",
                lines=8,
                interactive=False
            )
        ask_button.click(
            fn=answer_question,
            inputs=question_input,
            outputs=answer_output
        )

    gr.Markdown("<p style='text-align: center; color: gray;'>Made with ❤ using Gradio + Groq + LLaMA 3</p>")

if __name__ == "__main__":
  demo.launch()