import gradio as gr
from transformers import pipeline

# Load model
classifier = pipeline("sentiment-analysis", model="textattack/bert-base-uncased-imdb")

def predict(text):
    if not text.strip():
        return "⚠️ Please enter a review.", "", 0

    result = classifier(text)[0]
    label = result["label"].upper()
    score = round(result["score"], 4)

    # Map labels properly
    
    if label in ["POSITIVE", "LABEL_1"]:
        sentiment = "Positive"
        emoji = "😊"
    else:
        sentiment = "Negative"
        emoji = "😡"

    return f"{emoji} {sentiment}", f"Confidence: {score}", score


with gr.Blocks(theme=gr.themes.Soft(), title="IMDb Sentiment Analyzer") as demo:

    gr.Markdown(
        """
        # 🎬 IMDb Sentiment Analysis (BERT)
        Analyze movie reviews using a fine-tuned BERT model.
        """
    )

    with gr.Row():
        with gr.Column():
            text_input = gr.Textbox(
                label="Enter Movie Review",
                placeholder="Type your review here...",
                lines=6
            )

            with gr.Row():
                submit_btn = gr.Button("Analyze", variant="primary")
                clear_btn = gr.Button("Clear")

        with gr.Column():
            output_label = gr.Textbox(label="Prediction")
            output_conf = gr.Textbox(label="Confidence")
            confidence_bar = gr.Slider(
                minimum=0,
                maximum=1,
                label="Confidence Score",
                interactive=False
            )

    gr.Examples(
        examples=[
            "This movie was absolutely amazing, I loved every moment!",
            "Worst film I have ever seen. Totally waste of time.",
            "The acting was decent but the story was boring.",
            "Brilliant direction and outstanding performances!"
        ],
        inputs=text_input
    )

    submit_btn.click(
        fn=predict,
        inputs=text_input,
        outputs=[output_label, output_conf, confidence_bar]
    )

    clear_btn.click(
        fn=lambda: ("", "", 0),
        inputs=[],
        outputs=[output_label, output_conf, confidence_bar]
    )

demo.launch()