'''from transformers import GPT2LMHeadModel, GPT2Tokenizer
import gradio as gr
import torch

# Load model and tokenizer
model = GPT2LMHeadModel.from_pretrained("gpt2")
tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
tokenizer.pad_token = tokenizer.eos_token

# Function to convert or transform input text
def convert_text(input_text, task_type, max_length=80, temperature=0.9, top_p=0.9):
    # Craft a prompt based on the selected conversion task
    if task_type == "Paraphrase":
        prompt = f"Paraphrase this: {input_text}\nRewritten:"
    elif task_type == "Explain Simply":
        prompt = f"Explain this in simple words: {input_text}\nExplanation:"
    elif task_type == "Make it Formal":
        prompt = f"Convert to formal tone: {input_text}\nFormal:"
    elif task_type == "Make it Creative":
        prompt = f"Rewrite this creatively: {input_text}\nCreative version:"
    else:
        prompt = input_text

    # Tokenize and generate
    inputs = tokenizer(prompt, return_tensors="pt")
    outputs = model.generate(
        inputs.input_ids,
        max_length=max_length,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,
        pad_token_id=tokenizer.eos_token_id
    )

    result = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return result.replace(prompt, "").strip()

# Gradio interface
with gr.Blocks(title="🌀 GPT-2 Text Converter") as app:
    gr.Markdown("## 🔁 GPT-2 Text Converter\nChoose a task to transform your input text.")
    
    with gr.Row():
        input_text = gr.Textbox(label="Input Text", placeholder="Type or paste your sentence...", lines=3)
        task_type = gr.Dropdown(
            choices=["Paraphrase", "Explain Simply", "Make it Formal", "Make it Creative"],
            value="Paraphrase",
            label="Conversion Type"
        )

    with gr.Row():
        max_len = gr.Slider(30, 150, value=80, step=10, label="Max Output Length")
        temperature = gr.Slider(0.1, 2.0, value=0.9, step=0.1, label="Creativity (Temperature)")
        top_p = gr.Slider(0.5, 1.0, value=0.9, step=0.05, label="Top-p Sampling")

    output_text = gr.Textbox(label="Converted Output", lines=6)
    convert_button = gr.Button("🔄 Convert Text")

    convert_button.click(
        convert_text,
        inputs=[input_text, task_type, max_len, temperature, top_p],
        outputs=output_text
    )

app.launch()'''



from transformers import MarianMTModel, MarianTokenizer
import gradio as gr

# Language mapping: display name -> model name
lang_to_model = {
    "Hindi": "Helsinki-NLP/opus-mt-en-hi",
    "Bengali": "Helsinki-NLP/opus-mt-en-bn",
    "Tamil": "Helsinki-NLP/opus-mt-en-ta",
    "Telugu": "Helsinki-NLP/opus-mt-en-te",
    "Gujarati": "Helsinki-NLP/opus-mt-en-gu",
    "Marathi": "Helsinki-NLP/opus-mt-en-mr",
    "Malayalam": "Helsinki-NLP/opus-mt-en-ml",
    "Punjabi": "Helsinki-NLP/opus-mt-en-pa",
    "Urdu": "Helsinki-NLP/opus-mt-en-ur"
}

# Default load
default_lang = "Hindi"
model_name = lang_to_model[default_lang]
tokenizer = MarianTokenizer.from_pretrained(model_name)
model = MarianMTModel.from_pretrained(model_name)

# Translate function
def translate_text(text, target_lang):
    global tokenizer, model

    # Load model for selected language (if changed)
    selected_model = lang_to_model[target_lang]
    if tokenizer.name_or_path != selected_model:
        tokenizer = MarianTokenizer.from_pretrained(selected_model)
        model = MarianMTModel.from_pretrained(selected_model)

    # Translate
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
    translated = model.generate(**inputs, max_length=256)
    return tokenizer.decode(translated[0], skip_special_tokens=True)

# Gradio interface
with gr.Blocks(title="🌐 Indian Language Translator") as app:
    gr.Markdown("## 🌐 English to Indian Languages Translator")
    gr.Markdown("Type something in English and convert it into your selected Indian language.")

    input_text = gr.Textbox(label="Input (English)", lines=3, placeholder="Enter text in English")

    target_lang = gr.Dropdown(choices=list(lang_to_model.keys()), value=default_lang, label="Select Language")

    output_text = gr.Textbox(label="Translated Output", lines=3)

    translate_btn = gr.Button("🔄 Translate")

    translate_btn.click(translate_text, inputs=[input_text, target_lang], outputs=output_text)

app.launch()