app.py

import os
import gradio as gr
import torch
import logging
from langchain.chains import ConversationalRetrievalChain
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain.document_loaders import PyMuPDFLoader  # ✅ More stable PDF loader
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_community.vectorstores import Chroma
from langchain_community.llms import HuggingFacePipeline
from transformers import pipeline

# Setup Logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Set Hugging Face Cache Directory
os.environ["HF_HOME"] = "/tmp/huggingface_cache"

# Check for GPU availability
DEVICE = "cuda:0" if torch.cuda.is_available() else "cpu"

# Global variables
conversation_retrieval_chain = None
chat_history = []
llm_pipeline = None
embeddings = None
persist_directory = "/tmp/chroma_db"  # Storage for vector DB

def init_llm():
    """Initialize LLM and Embeddings"""
    global llm_pipeline, embeddings
    
    hf_token = os.getenv("HUGGINGFACEHUB_API_TOKEN")
    if not hf_token:
        raise ValueError("HUGGINGFACEHUB_API_TOKEN is not set in environment variables.")
    
    model_id = "tiiuae/falcon-rw-1b"  # ✅ Can switch to "tiiuae/falcon-rw-1b" for lighter model
    hf_pipeline = pipeline("text-generation", model=model_id, device=DEVICE)
    llm_pipeline = HuggingFacePipeline(pipeline=hf_pipeline)
    
    embeddings = HuggingFaceEmbeddings(
        model_name="sentence-transformers/all-MiniLM-L6-v2",
        model_kwargs={"device": DEVICE}
    )
    logger.info("✅ LLM and Embeddings Initialized Successfully!")

def process_document(file):
    """Process uploaded PDF and create a retriever"""
    global conversation_retrieval_chain
    
    if not llm_pipeline or not embeddings:
        init_llm()
    
    try:
        file_path = file.name  # ✅ Ensures correct file path is passed
        logger.info(f"📂 Processing PDF: {file_path}")

        loader = PyMuPDFLoader(file_path)  # ✅ Alternative loader for stability
        documents = loader.load()
        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1024, chunk_overlap=64)
        texts = text_splitter.split_documents(documents)

        # Load or create ChromaDB
        db = Chroma.from_documents(texts, embedding=embeddings, persist_directory=persist_directory)
        retriever = db.as_retriever(search_type="similarity", search_kwargs={'k': 6})

        conversation_retrieval_chain = ConversationalRetrievalChain.from_llm(
            llm=llm_pipeline, retriever=retriever
        )
        logger.info("✅ PDF Processed Successfully!")
        return "📄 PDF uploaded and processed successfully! You can now ask questions."
    
    except Exception as e:
        logger.error(f"❌ Error processing PDF: {str(e)}")
        return f"❌ Error processing PDF: {str(e)}"

def process_prompt(prompt, chat_history_display):
    """Generate a response using the retrieval chain"""
    global conversation_retrieval_chain, chat_history

    if not conversation_retrieval_chain:
        return chat_history_display + [("❌ No document uploaded.", "Please upload a PDF first.")]

    output = conversation_retrieval_chain({"question": prompt, "chat_history": chat_history})
    answer = output["answer"]

    chat_history.append((prompt, answer))
    return chat_history

# Define Gradio UI
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown("<h1 style='text-align: center;'>Personal Data Assistant</h1>")
    
    with gr.Row():
        dark_mode = gr.Checkbox(label="🌙 Toggle light/dark mode")
    
    with gr.Column():
        gr.Markdown("Hello there! I'm your friendly data assistant, ready to answer any questions regarding your data. Could you please upload a PDF file for me to analyze?")
        file_input = gr.File(label="Upload File")
        upload_button = gr.Button("📂 Upload File")
    
    status_output = gr.Textbox(label="Status", interactive=False)
    chat_history_display = gr.Chatbot(label="Chat History")
    
    with gr.Row():
        user_input = gr.Textbox(placeholder="Type your message here...", scale=4)
        submit_button = gr.Button("📩", scale=1)
        clear_button = gr.Button("🔄", scale=1)

    # Button Click Actions
    upload_button.click(process_document, inputs=file_input, outputs=status_output)
    submit_button.click(process_prompt, inputs=[user_input, chat_history_display], outputs=chat_history_display)
    clear_button.click(lambda: [], outputs=chat_history_display)

# Launch Gradio App
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)  # ✅ Works in Hugging Face Spaces