import gradio as gr
from transformers import pipeline

# Load the sentiment analysis pipeline
# We use a model specifically trained on product reviews (Amazon reviews)
model_name = "LiYuan/amazon-review-sentiment-analysis"
sentiment_pipeline = pipeline("sentiment-analysis", model=model_name)

def analyze_sentiment(review_text):
    """
    Analyzes the sentiment of the input text and returns a formatted result.
    The model outputs star ratings (1-5 stars).
    """
    if not review_text.strip():
        return "Please enter some text to analyze.", None

    try:
        # Perform sentiment analysis
        results = sentiment_pipeline(review_text)
        
        # The model returns labels like '1 star', '2 stars', etc.
        label = results[0]['label']
        score = results[0]['score']
        
        # Map star ratings to sentiment categories
        star_count = int(label.split()[0])
        
        if star_count >= 4:
            sentiment = "Positive"
            color = "🟢"
        elif star_count == 3:
            sentiment = "Neutral"
            color = "🟡"
        else:
            sentiment = "Negative"
            color = "🔴"
            
        result_text = f"### Sentiment: {sentiment} {color}\n"
        result_text += f"**Rating:** {label} ({score:.2%} confidence)\n\n"
        
        # Add some context for computer system products
        if "battery" in review_text.lower():
            result_text += "- *Note: This review mentions battery life.*\n"
        if "performance" in review_text.lower() or "fast" in review_text.lower() or "slow" in review_text.lower():
            result_text += "- *Note: This review mentions system performance.*\n"
        if "screen" in review_text.lower() or "display" in review_text.lower():
            result_text += "- *Note: This review mentions the display/screen.*\n"
            
        return result_text, {label: score}
        
    except Exception as e:
        return f"Error during analysis: {str(e)}", None

# Define the Gradio interface
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown("# 💻 Computer System Sentiment Analyzer")
    gr.Markdown(
        "Enter a review for a computer, laptop, or hardware component to analyze its sentiment. "
        "This tool uses a model trained on millions of product reviews to provide accurate star ratings."
    )
    
    with gr.Row():
        with gr.Column():
            input_text = gr.Textbox(
                label="Product Review",
                placeholder="e.g., The MacBook Pro has amazing performance and a stunning display, but the price is a bit high...",
                lines=5
            )
            submit_btn = gr.Button("Analyze Sentiment", variant="primary")
            
        with gr.Column():
            output_markdown = gr.Markdown(label="Analysis Result")
            output_label = gr.Label(label="Confidence Score")

    # Examples for users to try
    gr.Examples(
        examples=[
            ["The laptop is incredibly fast and the battery lasts all day. Highly recommended!"],
            ["The screen arrived with dead pixels and the customer service was unhelpful. Disappointed."],
            ["It's a decent computer for the price. Not the fastest, but gets the job done for basic tasks."],
            ["The cooling system is quite loud under load, but the gaming performance is top-notch."]
        ],
        inputs=input_text
    )

    submit_btn.click(
        fn=analyze_sentiment,
        inputs=input_text,
        outputs=[output_markdown, output_label]
    )

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