import gradio as gr
import pdfplumber
import together
import re
import unicodedata

# Set up Together.AI API Key (Replace with your actual key)
together.api_key = "8052996318422f1b9470840fc6ebc94e80676391f07e71cc15951b08bb430240"

def clean_text(text):
    """Cleans extracted text for better processing by the model."""
    print("cleaning")
    text = unicodedata.normalize("NFKC", text)  # Normalize Unicode characters
    text = re.sub(r'\s+', ' ', text).strip()  # Remove extra spaces and newlines
    text = re.sub(r'[^a-zA-Z0-9.,!?;:\'\"()\-]', ' ', text)  # Keep basic punctuation
    text = re.sub(r'(?i)(page\s*\d+)', '', text)  # Remove page numbers
    return text

def extract_text_from_pdf(pdf_file):
    """Extract and clean text from the uploaded PDF."""
    print("extracting")
    try:
        with pdfplumber.open(pdf_file) as pdf:
            text = " ".join(clean_text(text) for page in pdf.pages if (text := page.extract_text()))
        return text
    except Exception as e:
        print(f"Error extracting text: {e}")
        return None

def split_text(text, chunk_size=500):
    """Splits text into smaller chunks for better processing."""
    print("splitting")
    return [text[i:i+chunk_size] for i in range(0, len(text), chunk_size)]

def chatbot(pdf_file, user_question):
    """Processes the PDF and answers the user's question."""
    print("chatbot start")
    
    # Extract text from the PDF
    text = extract_text_from_pdf(pdf_file)
    if not text:
        return "Could not extract any text from the PDF."
    
    # Split into smaller chunks
    chunks = split_text(text)

    # Use only the first chunk (to optimize token usage)
    prompt = f"Based on this document, answer the question:\n\nDocument:\n{chunks[0]}\n\nQuestion: {user_question}"

    # Send to Together.AI (Mistral-7B)
    response = together.Completion.create(
        model="mistralai/Mistral-7B-Instruct-v0.1",
        prompt=prompt,
        max_tokens=200,
        temperature=0.7,
    )

    # Return chatbot's response
    return response.choices[0].text


# Gradio Interface
iface = gr.Interface(
    fn=chatbot,
    inputs=[gr.File(label="Upload PDF"), gr.Textbox(label="Ask a Question")],
    outputs=gr.Textbox(label="Answer"),
    title="PDF Q&A Chatbot (Powered by Together.AI)"
)

# Launch Gradio app
iface.launch()