app.py

"""
Myanmar Ghost - Gradio Demo for HuggingFace Spaces

This app provides an interactive demo for Myanmar sentiment analysis.
"""

import gradio as gr
import sys
from pathlib import Path

# Add src to path
sys.path.insert(0, str(Path(__file__).parent))

# Try to import the model components
try:
    from src.data_processing.text_normalizer import MyanmarTextNormalizer
    from src.augmentation.synonym_replacer import MyanmarSynonymReplacer
    MODULES_LOADED = True
except ImportError as e:
    MODULES_LOADED = False
    print(f"Warning: Some modules not loaded: {e}")


# Sentiment labels
SENTIMENT_LABELS = ["negative", "neutral", "positive", "sarcastic"]

# Initialize components
normalizer = MyanmarTextNormalizer() if MODULES_LOADED else None
replacer = MyanmarSynonymReplacer() if MODULES_LOADED else None


def analyze_sentiment(text: str) -> dict:
    """Analyze sentiment of Myanmar text."""
    if not text.strip():
        return {
            "text": text,
            "sentiment": "neutral",
            "confidence": 0.0,
            "probabilities": {label: 0.25 for label in SENTIMENT_LABELS},
        }
    
    # Normalize text
    if normalizer:
        normalized = normalizer.normalize_line(text)
    else:
        normalized = text
    
    # Mock prediction (replace with actual model)
    # In production, load the trained model here
    import random
    probs = [random.random() for _ in SENTIMENT_LABELS]
    total = sum(probs)
    probs = [p/total for p in probs]
    
    pred_idx = probs.index(max(probs))
    
    return {
        "text": text,
        "normalized_text": normalized if normalizer else text,
        "sentiment": SENTIMENT_LABELS[pred_idx],
        "confidence": max(probs),
        "probabilities": {SENTIMENT_LABELS[i]: probs[i] for i in range(4)},
    }


def get_synonyms(text: str) -> str:
    """Get synonym replacements for text."""
    if not replacer or not text.strip():
        return "Enter Myanmar text to see synonym examples"
    
    aug_text, replacements = replacer.augment_text(text, replace_prob=0.5)
    
    if not replacements:
        return f"No synonym replacements found for: {text}"
    
    result = f"Original: {text}\n\nAugmented: {aug_text}\n\nReplacements:\n"
    for orig, new in replacements:
        result += f"  • {orig} → {new}\n"
    
    return result


def show_probabilities(text: str) -> dict:
    """Show probability distribution."""
    result = analyze_sentiment(text)
    return result["probabilities"]


# Create Gradio interface
with gr.Blocks(
    title="Myanmar Ghost - Sentiment Analysis",
    theme=gr.themes.Soft(),
) as demo:
    
    gr.Markdown("""
    # 🇲🇲 Myanmar Ghost
    ### Advanced Myanmar Sentiment Analysis
    
    Enter Myanmar text to analyze sentiment. Supports:
    - ✅ Positive/Negative/Neutral/Sarcastic classification
    - 🔤 Text normalization
    - 📝 Synonym augmentation
    """)
    
    with gr.Row():
        with gr.Column(scale=2):
            text_input = gr.Textbox(
                label="Myanmar Text Input",
                placeholder="ကျေးဇူးပါ မင်္ဂလာပါ...",
                lines=3,
            )
            
            with gr.Row():
                analyze_btn = gr.Button("🔍 Analyze", variant="primary")
                clear_btn = gr.Button("🗑️ Clear")
        
        with gr.Column(scale=1):
            sentiment_output = gr.Label(
                label="Predicted Sentiment",
            )
    
    # Probabilities
    gr.Markdown("### 📊 Confidence Scores")
    prob_display = gr.BarPlot(
        x=["negative", "neutral", "positive", "sarcastic"],
        y=[0.25, 0.25, 0.25, 0.25],
        label="Probability Distribution",
        y_lab="Probability",
        x_lab="Sentiment",
    )
    
    # Synonym tool
    gr.Markdown("### 📝 Synonym Augmentation")
    with gr.Row():
        synonym_input = gr.Textbox(
            label="Text for Synonyms",
            placeholder="Enter text to see synonym replacements...",
            lines=2,
        )
        synonym_btn = gr.Button("🔄 Get Synonyms")
    
    synonym_output = gr.Textbox(
        label="Synonym Results",
        lines=4,
    )
    
    # Examples
    gr.Examples(
        examples=[
            ["ကျေးဇူးပါ"],
            ["မင်္ဂလာပါ"],
            ["မကျေနပ်ပါဗျ"],
            ["အရမ်းကောင်းတယ်"],
        ],
        inputs=text_input,
    )
    
    # Event handlers
    analyze_btn.click(
        fn=analyze_sentiment,
        inputs=text_input,
        outputs=[sentiment_output, prob_display],
    )
    
    clear_btn.click(
        fn=lambda: ("", {"negative": 0.25, "neutral": 0.25, "positive": 0.25, "sarcastic": 0.25}),
        inputs=[],
        outputs=[text_input, sentiment_output],
    )
    
    synonym_btn.click(
        fn=get_synonyms,
        inputs=synonym_input,
        outputs=synonym_output,
    )
    
    gr.Markdown("""
    ---
    ### ℹ️ About
    
    Myanmar Ghost is an advanced NLP project for Myanmar language understanding.
    
    - **Model**: Transformer-based sentiment classifier
    - **Features**: Multi-modal fusion, Active Learning, XAI
    - **Author**: [Aung Myo Kyaw](https://huggingface.co/amkyawdev)
    
    [GitHub Repository](https://github.com/amkyawdev/myanmar-ghost)
    """)


if __name__ == "__main__":
    demo.launch()