import os
import torch
import gradio as gr
import time
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
from code_flores_latest import flores_codes_latest


def load_models():
    model_name_dict = {'nllb-distilled-600M': 'facebook/nllb-200-distilled-600M'}
    model_dict = {}
    for call_name, real_name in model_name_dict.items():
        print('\tLoading model: %s' % call_name)
        model = AutoModelForSeq2SeqLM.from_pretrained(real_name)
        tokenizer = AutoTokenizer.from_pretrained(real_name)
        model_dict[call_name + '_model'] = model
        model_dict[call_name + '_tokenizer'] = tokenizer
    return model_dict


def translation(source, target, text):
    model_name = 'nllb-distilled-600M'
    source_code = flores_codes_latest.get(source, None)
    target_code = flores_codes_latest.get(target, None)
    if not source_code or not target_code:
        return "<p>Error: Language code not found.</p>"

    model = model_dict[model_name + '_model']
    tokenizer = model_dict[model_name + '_tokenizer']
    translator = pipeline('translation', model=model, tokenizer=tokenizer, src_lang=source_code, tgt_lang=target_code)
    output = translator(text, max_length=400)
    output_text = output[0]['translation_text']

    formatted_output = f"<p><strong>Original Text ({source}):</strong><br>{text}</p><p><strong>Translated Text ({target}):</strong> <span style='color: red;'>{output_text}</span></p>"
    return formatted_output


if __name__ == '__main__':
    print('\tInitializing models')
    model_dict = load_models()
    lang_names = list(flores_codes_latest.keys())
    source_dropdown = gr.Dropdown(lang_names, label='Source', allow_custom_value=True)
    target_dropdown = gr.Dropdown(lang_names, label='Target', allow_custom_value=True)
    textbox = gr.Textbox(lines=5, label="Input text")
    title = "nllb-distilled-600M - Example implementation"

    description = f"Nots: please note that not all translations are accurate, and some models codes are not accepted . "

    initial_output_value = "<p>Translation results will appear here.</p>"
    output_html = gr.HTML(label="Translation Result", value=initial_output_value)

    iface = gr.Interface(fn=translation, inputs=[source_dropdown, target_dropdown, textbox], outputs=output_html,
                         title=title, description=description)
    iface.launch()