import streamlit as st
import transformers
from transformers import pipeline
from transformers import AutoTokenizer, AutoModelForMaskedLM

tokenizer = AutoTokenizer.from_pretrained("moussaKam/AraBART", padding= True, truncation=True, max_length=128)
@st.cache
def load_model(model_name):
    model = AutoModelForMaskedLM.from_pretrained(model_name)
    return model
model = load_model("moussaKam/AraBART")


@st.cache
def next_word(text, pipe):
    res_dict= {  
  'token_str':[],
  'score':[],
  }
    res=pipe(text)
    for e in res:
        res_dict['token_str'].extend(e['token_str'])
        res_dict['score'].extend(e['score'])
    return res_dict

st.title("Predict Next Word")
st.write("Use our model to expand your query based on the DB content")
default_value = "التاريخ هو تحليل و"
# sent is the the variable holding the user's input
sent = st.text_area("Text", default_value, height = 60)
sent += ' <mask>'

pipe = pipeline("fill-mask", tokenizer = tokenizer, model = model)
dict_next_words = next_word(sent, pipe)

st.write(dict_next_words)