import gradio as gr
from collections import OrderedDict
from transformers import pipeline
def get_all_source_languages():
"""
Returns a human-readable `dict source_languages_names:codes`
based on the available models.
"""
source_languages = {
"English": "en",
"French": "fr",
"Spanish": "es",
"German": "de",
"Italian": "it"
}
source_languages = OrderedDict(sorted(source_languages.items()))
return source_languages
source_lang_dict = get_all_source_languages()
target_lang_dict = source_lang_dict # Reutilizamos el mismo diccionario para los idiomas de destino
def translate(input_text, source, target):
# Convertir nombres legibles a códigos abreviados
source_code = source_lang_dict.get(source, source) # 'en', 'fr', etc.
target_code = target_lang_dict.get(target, target) # 'en', 'fr', etc.
try:
model_name = f"Helsinki-NLP/opus-mt-{source_code}-{target_code}"
pipe = pipeline("translation", model=model_name)
translation = pipe(input_text)
return translation[0]['translation_text'], ""
except Exception as e:
return "", f"Error: {str(e)}"
with gr.Blocks() as demo:
gr.HTML("""
Open Translate
""")
with gr.Row():
with gr.Column():
source_language_dropdown = gr.Dropdown(
choices=list(source_lang_dict.keys()),
value=list(source_lang_dict.keys())[0],
label="Source Language"
)
input_textbox = gr.Textbox(
lines=5, placeholder="Enter text to translate", label="Input Text"
)
with gr.Column():
target_language_dropdown = gr.Dropdown(
choices=list(target_lang_dict.keys()),
value="English",
label="Target Language"
)
translated_textbox = gr.Textbox(lines=5, placeholder="", label="Translated Text")
info_label = gr.HTML("")
btn = gr.Button("Translate")
btn.click(
translate,
inputs=[input_textbox, source_language_dropdown, target_language_dropdown],
outputs=[translated_textbox, info_label]
)
if __name__ == "__main__":
demo.launch()