app.py

import streamlit as st
from transformers import MarianMTModel, MarianTokenizer
from PIL import Image

# Define available languages with MarianMT models
LANGUAGES = {
    'Spanish': 'es',
    'French': 'fr',
    'German': 'de',
    'Chinese': 'zh',
    'Hindi': 'hi',
    'Arabic': 'ar',
    'Japanese': 'ja',
    'Russian': 'ru',
    'Italian': 'it',
    'Portuguese': 'pt',
}

# Load a background image for the app
def add_bg_image(image_path):
    with open(image_path, "rb") as f:
        data = f.read()
    st.markdown(
        f"""
        <style>
        .stApp {{
            background-image: url(data:image/{"png"};base64,{data.encode("base64").decode()});
            background-size: cover;
        }}
        </style>
        """,
        unsafe_allow_html=True
    )

# Function to load the model based on the selected language
@st.cache_resource
def load_model(src_lang='en', tgt_lang='es'):
    model_name = f'Helsinki-NLP/opus-mt-{src_lang}-{tgt_lang}'
    model = MarianMTModel.from_pretrained(model_name)
    tokenizer = MarianTokenizer.from_pretrained(model_name)
    return model, tokenizer

# Function to translate text
def translate_text(model, tokenizer, text):
    inputs = tokenizer.encode(text, return_tensors='pt', truncation=True, padding=True)
    translated = model.generate(inputs, max_length=512, num_beams=5, early_stopping=True)
    translated_text = tokenizer.decode(translated[0], skip_special_tokens=True)
    return translated_text

# Load the background image
add_bg_image("path_to_your_image.png")  # You need to provide your image

# Streamlit app layout
st.markdown("<h1 style='text-align: center; color: #4B9CD3;'>🌍 Multilingual Translator 🌍</h1>", unsafe_allow_html=True)
st.markdown("<p style='text-align: center; color: #333; font-size: 18px;'>Translate English text into multiple languages</p>", unsafe_allow_html=True)

# Sidebar for language selection and instructions
st.sidebar.title("Language Options")
language = st.sidebar.selectbox("Choose target language", list(LANGUAGES.keys()))

st.sidebar.markdown("### How to use")
st.sidebar.write("1. Enter English text in the box below.")
st.sidebar.write("2. Select the target language from the options.")
st.sidebar.write("3. Click **Translate** to get the result.")

# Input text
text = st.text_area("Enter text in English to translate:", height=150)

# Character count
st.write(f"Character count: {len(text)}")

# Button to translate
if st.button("Translate"):
    if text:
        # Show a spinner during the translation process
        with st.spinner('Translating...'):
            # Load model and tokenizer based on selected language
            tgt_lang = LANGUAGES[language]
            model, tokenizer = load_model('en', tgt_lang)
            
            # Perform translation
            translated_text = translate_text(model, tokenizer, text)
        
        # Display the translation
        st.markdown("<h3 style='color: #4B9CD3;'>Translated Text:</h3>", unsafe_allow_html=True)
        st.success(translated_text)
    else:
        st.error("Please enter text to translate.")

# Footer with styling
st.markdown(
    """
    <style>
    footer {visibility: hidden;}
    .footer-text {
        position: fixed;
        bottom: 0;
        left: 0;
        right: 0;
        background-color: #f9f9f9;
        padding: 10px;
        text-align: center;
        color: #4B9CD3;
        font-weight: bold;
    }
    </style>
    <div class="footer-text">Powered by Hugging Face Transformers</div>
    """,
    unsafe_allow_html=True
)