import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
import torch

LANGS = ["kin_Latn","eng_Latn"]
TASK = "translation"
# CKPT = "DigitalUmuganda/Finetuned-NLLB"
MODELS = ["facebook/nllb-200-distilled-600M","DigitalUmuganda/Finetuned-NLLB"]
# model = AutoModelForSeq2SeqLM.from_pretrained(CKPT)
# tokenizer = AutoTokenizer.from_pretrained(CKPT)

device = 0 if torch.cuda.is_available() else -1

du_model = AutoModelForSeq2SeqLM.from_pretrained("DigitalUmuganda/Finetuned-NLLB")
fb_model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-distilled-600M")
models = {"facebook/nllb-200-distilled-600M":fb_model,"DigitalUmuganda/Finetuned-NLLB":du_model}
tokenizer = AutoTokenizer.from_pretrained("facebook/nllb-200-distilled-600M")
# def translate(text, src_lang, tgt_lang, max_length=400):
def translate(CKPT,text, src_lang, tgt_lang, max_length=400):

    """
    Translate the text from source lang to target lang
    """


    translation_pipeline = pipeline(TASK,
                                    model=models[CKPT],
                                    tokenizer=tokenizer,
                                    src_lang=src_lang,
                                    tgt_lang=tgt_lang,
                                    max_length=max_length,
                                    device=device)

    result = translation_pipeline(text)
    return result[0]['translation_text']


gr.Interface(
    translate,
    [
        gr.components.Dropdown(label="choose a model",choices=MODELS),
        gr.components.Textbox(label="Text"),
        gr.components.Dropdown(label="Source Language", choices=LANGS),
        gr.components.Dropdown(label="Target Language", choices=LANGS),
        #gr.components.Slider(8, 512, value=400, step=8, label="Max Length")
    ],
    ["text"],
    #examples=examples,
    # article=article,
    cache_examples=False,
    title="Finetuned-NLLB-1",
    #description=description
).launch()