app.py

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

# ==========================
# 1. Load model from Hugging Face
# ==========================

MODEL_NAME = "angkor96/khmer-news-summarization"  # e.g., "Sedtha-019/khmer-summarization"

print("Loading model and tokenizer...")
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME)

# Move to GPU if available
device = "cuda" if torch.cuda.is_available() else "cpu"
model = model.to(device)

print(f"✅ Model loaded successfully on {device}!")

# ==========================
# 2. Summarization function
# ==========================
def summarize_khmer_text(text, max_length=150):
    """
    Summarize Khmer text
    """
    if not text or text.strip() == "":
        return "⚠️ សូមបញ្ចូលអត្ថបទ / Please enter text"
    
    # if len(text.strip()) < 20:
    #     return "⚠️ អត្ថបទខ្លីពេក / Text is too short to summarize"
    
    try:
        # Tokenize input
        inputs = tokenizer(
            text,
            max_length=1024,
            truncation=True,
            padding="max_length",
            return_tensors="pt"
        ).to(device)
        
        # Generate summary
        with torch.no_grad():
            summary_ids = model.generate(
                inputs["input_ids"],
                max_length=max_length,
                # min_length=min_length,
                length_penalty=2.0,
                num_beams=5,
                early_stopping=True,
                # no_repeat_ngram_size=3
            )
        
        # Decode output
        summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
        
        return summary
        
    except Exception as e:
        return f"❌ Error: {str(e)}"

# ==========================
# 3. Gradio UI
# ==========================
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown(
        """
        # 🇰🇭 Khmer Text Summarization
        ### បញ្ចូលអត្ថបទខ្មែរ ហើយទទួលបានការសង្ខេបដោយស្វ័យប្រវត្តិ
        Enter Khmer text and get an automatic summary
        """
    )
    
    with gr.Row():
        with gr.Column():
            input_text = gr.Textbox(
                lines=10,
                placeholder="បញ្ចូលអត្ថបទខ្មែរនៅទីនេះ...\nEnter Khmer text here...",
                label="📝 អត្ថបទដើម / Original Text"
            )
            
            with gr.Row():
                max_len = gr.Slider(
                    minimum=50,
                    maximum=300,
                    value=150,
                    step=10,
                    label="Maximum Summary Length"
                )
                min_len = gr.Slider(
                    minimum=20,
                    maximum=100,
                    value=40,
                    step=10,
                    label="Minimum Summary Length"
                )
            
            submit_btn = gr.Button("🔄 Summarize / សង្ខេប", variant="primary")
        
        with gr.Column():
            output_text = gr.Textbox(
                lines=10,
                label="📋 សង្ខេប / Summary"
            )
    
    # Examples
    gr.Examples(
        examples=[
            ["ប្រទេសកម្ពុជាមានប្រវត្តិសាស្រ្តយូរលង់និងសម្បូរបែបដោយវប្បធម៌។ អាណាចក្រខ្មែរបានរីកចម្រើនក្នុងសតវត្សទី៩ដល់ទី១៥។ អង្គរវត្តជាស្នាដៃស្ថាបត្យកម្មដ៏អស្ចារ្យមួយរបស់ពិភពលោក។", 100, 30],
            ["ការអប់រំជាមូលដ្ឋានគ្រឹះសំខាន់សម្រាប់ការអភិវឌ្ឍន៍ជាតិ។ សិស្សានុសិស្សគប្បីរៀនសូត្រយ៉ាងស្អិតរុំ។ គ្រូបង្រៀនមានតួនាទីសំខាន់ក្នុងការបង្កើតអនាគតកុមារ។", 80, 25],
        ],
        inputs=[input_text, max_len, min_len],
    )
    
    # Connect button
    submit_btn.click(
        fn=summarize_khmer_text,
        inputs=[input_text, max_len, min_len],
        outputs=output_text
    )

# ==========================
# 4. Launch
# ==========================
if __name__ == "__main__":
    demo.launch(share=True)