import gradio as gr
import requests
import random
from transformers import pipeline


classifier = pipeline("zero-shot-classification", model="valhalla/distilbart-mnli-12-3")

labels = [
    "animals", "people", "places", "history", "science", "art", "technology",
    "sports", "food", "clothing", "home", "entertainment", "education", "nature", "transportation"
]


def preprocess_topic(topic):
    topic = topic.lower().strip()
    mapping = {
        "shirt": "clothing item shirt",
        "jeans": "clothing item jeans",
        "shoes": "clothing item shoes",
        "dress": "clothing item dress",
        "sandals": "clothing item sandals",
        "cookie": "sweet snack cookie",
        "orcas": "marine mammal orcas",
        "penguin": "bird that swims penguin",
        "whale": "large marine animal whale",
        "floor": "interior surface floor",
        "blanket": "household item blanket",
        "bed": "furniture item bed",
        "lamp": "household lighting lamp",
        "girl": "young person female",
        "bag": "fashion accessory bag",
        "kids": "young people children",
    }
    return mapping.get(topic, topic)

random_topics = [
    "cats", "space", "chocolate", "Egypt", "Leonardo da Vinci",
    "volcanoes", "Tokyo", "honeybees", "quantum physics", "orcas"
]


def get_wikipedia_facts(topic):
    if not topic.strip():
        return "Please enter a topic or use 'Surprise me!'", None, None

    headers = {
        "User-Agent": "SmartFactApp/1.0"
    }

    params = {
        "action": "query",
        "format": "json",
        "prop": "extracts|pageimages",
        "exintro": True,
        "explaintext": True,
        "piprop": "thumbnail",
        "pithumbsize": 400,
        "generator": "search",
        "gsrsearch": topic,
        "gsrlimit": 1,
    }

    try:
        response = requests.get("https://en.wikipedia.org/w/api.php", params=params, headers=headers, timeout=5)
        data = response.json()

        pages = data.get("query", {}).get("pages", {})
        if not pages:
            return f"Sorry, no information found for '{topic}'.", None, None

        page = next(iter(pages.values()))
        extract_text = page.get("extract", "")
        image_url = page.get("thumbnail", {}).get("source", None)

        sentences = [s.strip() for s in extract_text.replace("\n", " ").split(". ") if s.strip()]
        facts = random.sample(sentences, min(3, len(sentences)))
        facts = [fact if fact.endswith(".") else fact + "." for fact in facts]
        facts_text = "\n\n".join(f"💡 {fact}" for fact in facts)

       
        processed_input = preprocess_topic(topic)
        classification = classifier(processed_input, candidate_labels=labels)
        top_labels = classification["labels"][:3]
        top_scores = classification["scores"][:3]

        classification_text = "🧠 Top categories:\n"
        if top_scores[0] < 0.3:
            classification_text = "🧠 Category uncertain.\n" + classification_text
        for label, score in zip(top_labels, top_scores):
            classification_text += f"- {label} ({score:.2%})\n"

        return facts_text, image_url, classification_text

    except Exception as e:
        print("Error:", e)
        return "Oops! Something went wrong while fetching your facts.", None, None


def surprise_topic(_):
    topic = random.choice(random_topics)
    return get_wikipedia_facts(topic)


with gr.Blocks() as demo:
    gr.HTML("""
    <style>
      body {
        background-color: #ADD8E6 !important;
      }
      .gradio-container {
        background-color: transparent !important;
      }
    </style>
    """)

    gr.Markdown("""
# 🌍 Smart Wikipedia Fact Finder  
Search any topic and discover:
- 📚 Three interesting facts
- 🖼️ A related image
- 🧠 AI-predicted topic category

👉 Try something like **"bed"**, **"quantum physics"**, or click **🎲 Surprise me!**
    """)

    with gr.Row():
        with gr.Column(scale=3):
            topic_input = gr.Textbox(label="Enter a Topic", placeholder="e.g. Eiffel Tower, cookies, World War II")
        with gr.Column(scale=1):
            surprise_button = gr.Button("🎲 Surprise me!")

    gr.Markdown("---")

    with gr.Row():
        with gr.Column():
            facts_output = gr.Textbox(label="📚 Wikipedia Facts", lines=6)
            classification_output = gr.Textbox(label="🧠 Topic Classification")
        with gr.Column():
            image_output = gr.Image(label="🖼️ Related Image")

    topic_input.submit(get_wikipedia_facts, inputs=topic_input, outputs=[facts_output, image_output, classification_output])
    surprise_button.click(surprise_topic, inputs=None, outputs=[facts_output, image_output, classification_output])

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