app.py

import gradio as gr
from transformers import pipeline
import urllib.parse

def extract_keywords(text: str, nlp_pipeline) -> list:
    """Extract relevant keywords from the input text"""
    prompt = f"Extract 3-4 most relevant gift-related keywords from: {text}\nKeywords:"
    response = nlp_pipeline(prompt, max_new_tokens=30, num_return_sequences=1)
    keywords = response[0]['generated_text'].split('Keywords:')[-1].strip()
    return [k.strip() for k in keywords.split(',') if k.strip()]

def generate_search_urls(keywords: list) -> dict:
    """Generate search URLs for various e-commerce platforms"""
    # Encode queries properly for URLs
    query = urllib.parse.quote(" ".join(keywords))

    return {
        "Amazon India": f'<a href="https://www.amazon.in/s?k={query}" target="_blank">Amazon</a>',
        "Flipkart": f'<a href="https://www.flipkart.com/search?q={query}" target="_blank">Flipkart</a>',
        "IGP Gifts": f'<a href="https://www.igp.com/search?q={query}" target="_blank">IGP</a>',
        "IndiaMart": f'<a href="https://www.indiamart.com/find?q={query}" target="_blank">IndiaMart</a>',
    }

def recommend_gifts(text: str):
    """Main function to generate gift recommendations"""
    if not text:
        return "⚠️ Please provide a description."

    try:
        # Load GPT-2 as a text-generation model
        nlp = pipeline(
            "text-generation",
            model="gpt2",
            device_map="auto"
        )

        # Extract relevant keywords
        keywords = extract_keywords(text, nlp)

        # Generate search URLs
        search_links = generate_search_urls(keywords)

        # Format the output as clickable links
        formatted_output = f"""
        <h3>🔍 Predicted Interests: {", ".join(keywords)}</h3>
        <h3>🛒 Gift Suggestions:</h3>
        <ul>
        <li>{search_links["Amazon India"]}</li>
        <li>{search_links["Flipkart"]}</li>
        <li>{search_links["IGP Gifts"]}</li>
        <li>{search_links["IndiaMart"]}</li>
        </ul>
        """
        return formatted_output

    except Exception as e:
        return f"❌ Error: {str(e)}"

# Create Gradio interface with HTML output
demo = gr.Interface(
    fn=recommend_gifts,
    inputs=gr.Textbox(
        lines=3,
        placeholder="Describe who you're buying a gift for (age, interests, occasion, etc.)"
    ),
    outputs=gr.HTML(),  # Change output type to HTML
    title="🎁 Smart Gift Recommender",
    description="Get personalized gift suggestions with direct shopping links!",
    examples=[
        ["a small kid of age 3 want him to have something like a toy that teaches alphabets"],
        ["age is 25 and he loves puzzle and online FPS games"],
        ["Looking for a gift for my mom who enjoys gardening and cooking"]
    ]
)

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