import gradio as gr
from detector import CustomDetector
import logging

# Set up logging to capture detailed errors
logging.basicConfig(level=logging.ERROR)
logger = logging.getLogger(__name__)

# Initialize the detector
try:
    detector = CustomDetector()
    logger.info("Detector initialized successfully")
except Exception as e:
    logger.error(f"Failed to initialize detector: {str(e)}")
    raise RuntimeError(f"Failed to initialize detector: {str(e)}")

def detect_text(text):
    """
    Compute the AI-generated text score for the input text.
    Returns a string with the score and classification (AI-generated or human-generated).
    """
    if not text.strip():
        return "Please enter some text."
    
    try:
        logger.info(f"Processing text: {text[:50]}...")
        score = detector.my_detector([text])[0]  # Process single text
        classification = "AI-generated" if score > 0.2 else "human-generated"
        interpretation = (
            f"Score: {score:.4f}\n"
            f"Classification: {classification}\n"
            "Note: Scores above 0.2 indicate AI-generated text; scores below 0.2 indicate human-generated text."
        )
        logger.info(f"Result: {interpretation}")
        return interpretation
    except Exception as e:
        error_msg = f"Error: {str(e)}"
        logger.error(error_msg)
        return error_msg

# Minimal CSS for a clean, professional look
custom_css = """
/* General styling */
body {
    font-family: -apple-system, BlinkMacSystemFont, sans-serif;
    background: #f5f7fa;
    color: #1a1a1a;
    line-height: 1.6;
}

/* Flexible container */
.gradio-container {
    max-width: 95%;
    width: 600px;
    margin: 1rem auto;
    padding: 1rem;
    background: #ffffff;
    border-radius: 8px;
    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
}

/* Title */
h1 {
    font-size: 2rem;
    font-weight: 600;
    text-align: center;
    color: #2c3e50;
    margin-bottom: 1rem;
}

/* Input textbox */
textarea {
    border: 1px solid #d1d5db !important;
    border-radius: 6px !important;
    padding: 0.75rem !important;
    font-size: 1rem !important;
}

/* Output textbox */
.output-text {
    background: #f9fafb !important;
    border-radius: 6px !important;
    padding: 1rem !important;
    font-size: 1rem !important;
    border: 1px solid #d1d5db !important;
}

/* Button */
button {
    background: #4b5563 !important;
    color: white !important;
    border: none !important;
    padding: 0.5rem 1rem !important;
    border-radius: 6px !important;
    font-weight: 500 !important;
}

/* Accordion */
.gr-accordion {
    margin-top: 1rem;
    border: 1px solid #d1d5db;
    border-radius: 6px;
}

/* Responsive design */
@media (max-width: 600px) {
    .gradio-container {
        margin: 0.5rem;
        padding: 0.5rem;
        width: 98%;
    }
    h1 {
        font-size: 1.6rem;
    }
}
"""

# Citation for the expandable tab
citation_markdown = """
## Citation

Please cite our work as:

**Zero-Shot Statistical Tests for LLM-Generated Text Detection using Finite Sample Concentration Inequalities**  
Tara Radvand, Mojtaba Abdolmaleki, Mohamed Mostagir, Ambuj Tewari  
[arXiv:2501.02406](https://arxiv.org/abs/2501.02406)  
Year: 2025
"""

# Set up the Gradio interface
with gr.Blocks(css=custom_css, theme=None) as iface:
    gr.Markdown("# AI-Generated Text Detector")
    gr.Markdown("Enter text to detect if it was generated by an AI model. Powered by a custom detector using tiiuae/falcon-rw-1b.")
    gr.Textbox(
        lines=5,
        placeholder="Enter text here to check if it's AI-generated...",
        label="Input Text"
    )
    output = gr.Textbox(label="Detection Result")
    gr.Button("Detect").click(
        fn=detect_text,
        inputs=gr.Textbox(),
        outputs=output
    )
    gr.Accordion("Citation", open=False).markdown(citation_markdown)

# Launch the app
if __name__ == "__main__":
    iface.launch()