from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
import torch
import pickle
import streamlit as st
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")

# model_name = "MoritzLaurer/mDeBERTa-v3-base-mnli-xnli"
# tokenizer = AutoTokenizer.from_pretrained(model_name)
# model = AutoModelForSequenceClassification.from_pretrained(model_name)

classifier = pipeline("zero-shot-classification", model="MoritzLaurer/mDeBERTa-v3-base-mnli-xnli")

# with open('chapter_titles.pkl', 'rb') as file:
#     titles_astiko = pickle.load(file)
# labels1 = ["κληρονομικό", "ακίνητα", "διαζύγιο"]
# # labels2 = ["αποδοχή κληρονομιάς", "αποποίηση", "διαθήκη"]
# # labels3 = ["μίσθωση", "κυριότητα", "έξωση", "απλήρωτα νοίκια"]


# titles_astiko = ["γάμος", "αλλοδαπός", "φορολογία", "κληρονομικά", "στέγη", "οικογενειακό", "εμπορικό","κλοπή","απάτη"]
# Load dictionary from the file using pickle
with open('my_dict.pickle', 'rb') as file:
    dictionary = pickle.load(file)

def classify(text,labels):
    output = classifier(text, labels, multi_label=False)
    
    return output


text = st.text_input('Enter some text:')  # Input field for new text

if text:

    labels = list(dictionary)
    
    output = classify(text,labels)

    output = output["labels"][0]

    labels = list(dictionary[output])

    output2 = classify(text,labels)

    output2 = output2["labels"][0]


    answer = dictionary[output][output2]

    st.text(output)
    st.text(output2)

    st.text(answer)