app.py

import os
import gradio as gr
from dotenv import load_dotenv
from huggingface_hub import InferenceClient
from langchain_community.document_loaders import PyPDFLoader
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import FAISS

print(f"Gradio version: {gr.__version__}")

# Load env
load_dotenv()
HF_TOKEN = os.getenv("HF_TOKEN")
print("HF_TOKEN loaded:", bool(HF_TOKEN))

# Free HF model
client = InferenceClient(
    model="meta-llama/Llama-3.2-1B-Instruct",
    token=HF_TOKEN
)

# Global vector store
vector_store = None

def process_pdf(pdf_file):
    """Load and process PDF into vector store"""
    global vector_store
    
    if pdf_file is None:
        return "Please upload a PDF file."
    
    # Load PDF
    loader = PyPDFLoader(pdf_file.name)
    documents = loader.load()
    
    # Split into chunks
    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=500,
        chunk_overlap=50
    )
    chunks = text_splitter.split_documents(documents)
    
    # Create embeddings and vector store
    embeddings = HuggingFaceEmbeddings(
        model_name="sentence-transformers/all-MiniLM-L6-v2"
    )
    vector_store = FAISS.from_documents(chunks, embeddings)
    
    return f"✅ PDF processed! {len(chunks)} chunks created. You can now ask questions."

def get_bot_response(message, history):
    global vector_store
    
    context = ""
    
    # If PDF is loaded, retrieve relevant context
    if vector_store is not None:
        docs = vector_store.similarity_search(message, k=3)
        context = "\n\n".join([doc.page_content for doc in docs])
    
    # Build the prompt
    if context:
        system_prompt = f"""You are a helpful AI assistant. Answer questions based on the provided context.
If the answer is not in the context, say so.

Context:
{context}"""
    else:
        system_prompt = "You are a helpful AI assistant."
    
    messages = [
        {"role": "system", "content": system_prompt},
        {"role": "user", "content": message}
    ]

    response = client.chat.completions.create(
        messages=messages,
        max_tokens=500,
        temperature=0.7,
    )

    return response.choices[0].message.content

def clear_pdf():
    """Clear the uploaded PDF"""
    global vector_store
    vector_store = None
    return "PDF cleared. Chat is now in general mode."

# Gradio UI
with gr.Blocks(title="🤖 AI Chatbot with PDF") as demo:
    gr.Markdown("# 🤖 AI Chatbot with PDF Support")
    gr.Markdown("Upload a PDF to chat about its contents, or just chat normally.")
    
    with gr.Row():
        with gr.Column(scale=1):
            pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
            upload_btn = gr.Button("Process PDF", variant="primary")
            clear_btn = gr.Button("Clear PDF")
            status = gr.Textbox(label="Status", interactive=False)
        
        with gr.Column(scale=2):
            chatbot_display = gr.Chatbot(label="Chat", height=400)
            msg_input = gr.Textbox(label="Your message", placeholder="Type your message here...")
            clear_chat_btn = gr.Button("Clear Chat")
    
    # Chat history state - using messages format for Gradio 6.x
    chat_history = gr.State([])
    
    def respond(message, history):
        if not message:
            return "", history, history
        
        bot_response = get_bot_response(message, history)
        
        # Gradio 6.x requires messages format with role and content
        history = history + [
            gr.ChatMessage(role="user", content=message),
            gr.ChatMessage(role="assistant", content=bot_response)
        ]
        return "", history, history
    
    def clear_chat():
        return [], []
    
    msg_input.submit(respond, [msg_input, chat_history], [msg_input, chatbot_display, chat_history])
    clear_chat_btn.click(clear_chat, outputs=[chatbot_display, chat_history])
    upload_btn.click(process_pdf, inputs=[pdf_input], outputs=[status])
    clear_btn.click(clear_pdf, outputs=[status])

demo.launch()