app.py

import os
import gradio as gr
import requests
from PyPDF2 import PdfReader
from langchain.text_splitter import CharacterTextSplitter
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import FAISS

# Load Groq API key from Hugging Face secrets
GROQ_API_KEY = os.getenv("GROQ_API_KEY")

# Load embedding model
embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")

# Initialize FAISS DB (in-memory for each session)
vectorstore = None


def load_and_embed_pdfs(pdf_files):
    global vectorstore

    all_text = ""
    for file in pdf_files:
        pdf = PdfReader(file.name)
        for page in pdf.pages:
            all_text += page.extract_text() or ""

    # Split text
    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
    texts = text_splitter.split_text(all_text)

    # Embed and store in FAISS
    vectorstore = FAISS.from_texts(texts, embedding_model)

    return "PDFs processed and indexed. You can now ask questions."


def groq_chat_completion(messages):
    headers = {
        "Content-Type": "application/json",
        "Authorization": f"Bearer {GROQ_API_KEY}"
    }
    data = {
        "model": "llama3-70b-8192",  # Or llama-3.3-70b-versatile
        "messages": messages
    }
    response = requests.post("https://api.groq.com/openai/v1/chat/completions", headers=headers, json=data)
    response.raise_for_status()
    return response.json()["choices"][0]["message"]["content"]


def ask_question(user_query, chat_history):
    global vectorstore

    if vectorstore is None:
        return "Please upload and process PDF documents first."

    # Perform similarity search
    docs = vectorstore.similarity_search(user_query, k=3)
    context = "\n\n".join(doc.page_content for doc in docs)

    # Create chat message format
    messages = [
        {"role": "system", "content": "Answer only based on the provided document context."},
        {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {user_query}"}
    ]

    # Get response
    try:
        answer = groq_chat_completion(messages)
    except Exception as e:
        return f"Error: {str(e)}"

    chat_history.append((user_query, answer))
    return "", chat_history


# Gradio interface
with gr.Blocks() as demo:
    gr.Markdown("## 📄 Chat with your PDF (Powered by Groq + Llama3)")
    pdf_upload = gr.File(file_types=[".pdf"], file_count="multiple", label="Upload PDF(s)")
    process_btn = gr.Button("Process PDFs")

    status_output = gr.Textbox(label="Status", interactive=False)

    chatbot = gr.Chatbot(label="Ask Questions")
    query_input = gr.Textbox(label="Your Question")
    send_btn = gr.Button("Send")

    # Bind actions
    process_btn.click(load_and_embed_pdfs, inputs=pdf_upload, outputs=status_output)
    send_btn.click(fn=ask_question, inputs=[query_input, chatbot], outputs=[query_input, chatbot])

# Launch
demo.launch()