import torch
import streamlit as st
from transformers import PegasusForConditionalGeneration, AutoTokenizer

@st.cache(allow_output_mutation=True)
def do_summary(model_name):
    model = PegasusForConditionalGeneration.from_pretrained(model_name)
    return model

@st.cache(allow_output_mutation=True)
def do_tokenize(model_name):
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    return tokenizer

model = do_summary("google/pegasus-cnn_dailymail")
tokenizer = do_tokenize("google/pegasus-cnn_dailymail")


def summarize(passage):
    txt = " ".join(passage)
    #model_name = 'google/pegasus-cnn_dailymail'
    device = 'cuda' if torch.cuda.is_available() else 'cpu'
    #tokenizer = AutoTokenizer.from_pretrained(model_name)
    #model = PegasusForConditionalGeneration.from_pretrained(model_name).to(device)
    batch = tokenizer(txt, truncation=True, padding='longest', return_tensors="pt").to(device)
    translated = model.generate(**batch)
    summy = tokenizer.batch_decode(translated, skip_special_tokens=True)
    print("summ end")
    return summy