app.py

import gradio as gr
from transformers import pipeline, AutoTokenizer
from sentence_transformers import SentenceTransformer, util

# Translation models
translation_models = {
    'Vietnamese': "Helsinki-NLP/opus-mt-en-vi",
    'Japanese': "Helsinki-NLP/opus-mt-en-jap",
    'Thai': "Helsinki-NLP/opus-mt-en-tha",
    'Spanish': "Helsinki-NLP/opus-mt-en-es"
}

# Initialize summarization pipeline with a specified model
model_name = "sshleifer/distilbart-cnn-12-6"
summarizer = pipeline("summarization", model=model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

# Initialize translation pipeline
def get_translator(language):
    model_name = translation_models.get(language)
    if model_name:
        return pipeline("translation", model=model_name)
    return None

# Helper function to generate bullet points
def generate_bullet_points(text):
    model = SentenceTransformer('paraphrase-MiniLM-L6-v2')
    sentences = text.split('. ')
    embeddings = model.encode(sentences, convert_to_tensor=True)
    clusters = util.community_detection(embeddings, threshold=0.75)
    
    bullet_points = []
    for cluster in clusters:
        cluster_sentences = [sentences[idx] for idx in cluster]
        main_sentence = cluster_sentences[0] if cluster_sentences else ""
        bullet_points.append(main_sentence.strip())
    
    return "\n".join(f"- {point}" for point in bullet_points)

# Helper function to split text into chunks
def split_text(text, max_tokens=1024):
    inputs = tokenizer(text, return_tensors='pt', truncation=False)
    input_ids = inputs['input_ids'][0]
    total_tokens = len(input_ids)
    
    chunks = []
    start = 0
    while start < total_tokens:
        end = min(start + max_tokens, total_tokens)
        chunk_ids = input_ids[start:end]
        chunk_text = tokenizer.decode(chunk_ids, skip_special_tokens=True)
        chunks.append(chunk_text)
        start = end
    
    return chunks

# Helper function to summarize text
def summarize_text(text):
    chunks = split_text(text)
    summaries = [summarizer(chunk, max_length=150, min_length=40, do_sample=False)[0]['summary_text'] for chunk in chunks]
    return " ".join(summaries)

# Helper function to translate text
def translate_text(text, language):
    translator = get_translator(language)
    if translator:
        translated_text = translator(text)[0]['translation_text']
        return translated_text
    return text

def process_text(input_text, language):
    summary = summarize_text(input_text)
    bullet_points = generate_bullet_points(summary)
    translated_text = translate_text(bullet_points, language)
    return bullet_points, translated_text

# Create Gradio interface
iface = gr.Interface(
    fn=process_text,
    inputs=[
        gr.Textbox(label="Input Text", placeholder="Paste your text here...", lines=10),
        gr.Dropdown(choices=["Vietnamese", "Japanese", "Thai", "Spanish"], label="Translate to", value="Vietnamese")
    ],
    outputs=[
        gr.Textbox(label="Bullet Points", lines=10),
        gr.Textbox(label="Translated Bullet Points", lines=10)
    ],
    title="Text to Bullet Points and Translation",
    description="Paste any text, and the program will summarize it into bullet points. Optionally, translate the bullet points into Vietnamese, Japanese, Thai, or Spanish."
)

iface.launch()