app.py

import streamlit as st
from transformers import MarianMTModel, MarianTokenizer

# Function to load the translation model and tokenizer
@st.cache_resource
def load_model_and_tokenizer():
    model_name = "Helsinki-NLP/opus-mt-en-ROMANCE"  # Example: English to Romance languages (like Spanish, French, etc.)
    model = MarianMTModel.from_pretrained(model_name)
    tokenizer = MarianTokenizer.from_pretrained(model_name)
    return model, tokenizer

# Function to translate text
def translate_text(text, model, tokenizer, src_lang, tgt_lang):
    # Prepare the text for translation
    translation_input = tokenizer(text, return_tensors="pt", padding=True)
    
    # Translate text
    translated = model.generate(**translation_input)
    
    # Decode the translated text
    translated_text = tokenizer.decode(translated[0], skip_special_tokens=True)
    return translated_text

# Streamlit UI components
st.title("Language Translation App")
st.write("Translate text between multiple languages using the Helsinki-NLP translation models.")

# Load model and tokenizer
model, tokenizer = load_model_and_tokenizer()

# Language selection for input and output
available_languages = ["en", "es", "fr", "de", "it", "pt", "ro", "nl", "pl", "ca"]
input_lang = st.selectbox("Select input language", available_languages)
output_lang = st.selectbox("Select output language", available_languages)

# Text input
text_to_translate = st.text_area("Enter text to translate:")

# Perform translation if text is entered
if text_to_translate:
    if input_lang != output_lang:
        translated_text = translate_text(text_to_translate, model, tokenizer, input_lang, output_lang)
        st.write("Translated Text:")
        st.write(translated_text)
    else:
        st.warning("Input and output languages cannot be the same.")