app.py

import streamlit as st
import requests
import threading
from flask import Flask, request, jsonify
from transformers import AutoTokenizer, AutoModelForQuestionAnswering
import torch

# Modell betöltése
tokenizer = AutoTokenizer.from_pretrained("nlpaueb/legal-bert-base-uncased")
model = AutoModelForQuestionAnswering.from_pretrained("nlpaueb/legal-bert-base-uncased")

# Flask API létrehozása háttérfolyamatként
app = Flask(__name__)

@app.route("/answer", methods=["POST"])
def answer():
    data = request.json
    context = data.get("context")
    question = data.get("question")

    # Tokenizálás és válasz előállítás
    inputs = tokenizer.encode_plus(question, context, return_tensors="pt")
    answer_start_scores, answer_end_scores = model(**inputs).values()

    # Legjobb válasz kiválasztása
    answer_start = torch.argmax(answer_start_scores)
    answer_end = torch.argmax(answer_end_scores) + 1
    answer = tokenizer.convert_tokens_to_string(
        tokenizer.convert_ids_to_tokens(inputs["input_ids"][0][answer_start:answer_end])
    )

    return jsonify({"answer": answer})

def run_flask_app():
    app.run(port=5000)

# Flask szerver indítása háttérszálon
flask_thread = threading.Thread(target=run_flask_app)
flask_thread.start()

# Streamlit alkalmazás
def split_text_into_chunks(text, max_length=512):
    paragraphs = text.split("\n\n")
    chunks = []
    current_chunk = ""
    
    for paragraph in paragraphs:
        if len(current_chunk) + len(paragraph) <= max_length:
            current_chunk += paragraph + "\n\n"
        else:
            chunks.append(current_chunk)
            current_chunk = paragraph + "\n\n"
    
    if current_chunk:
        chunks.append(current_chunk)
    
    return chunks

st.title("AI Jogi Chatbot")

# Dokumentum feltöltése
uploaded_file = st.file_uploader("Töltsön fel egy dokumentumot", type=["txt", "pdf"])

if uploaded_file:
    context = uploaded_file.read().decode("utf-8")  # Szöveg kinyerése
    chunks = split_text_into_chunks(context)
    st.write(f"A dokumentum {len(chunks)} részre bontva.")
    
    # Felhasználói kérdés
    question = st.text_input("Írja be a kérdését a dokumentumról:")

    if question:
        answers = []
        for i, chunk in enumerate(chunks):
            response = requests.post(
                "http://localhost:5000/answer",
                json={"context": chunk, "question": question},
            )
            if response.status_code == 200:
                answer = response.json().get("answer")
                answers.append(f"Rész {i+1}: {answer}")
        
        st.write("Válaszok:")
        for ans in answers:
            st.write(ans)