import gradio as gr
import requests
import os
from openai import OpenAI

# Initialize Groq LLM client
client = OpenAI(
    api_key=os.getenv("GROQ_API_KEY"),
    base_url="https://api.groq.com/openai/v1",
)

# Store search results
product_cache = []

# Step 1: Search products

def search_products(product_name):
    global product_cache
    url = f"https://world.openfoodfacts.org/cgi/search.pl?search_terms={product_name}&search_simple=1&action=process&json=1&page_size=10"
    response = requests.get(url)
    if response.status_code != 200:
        return [], gr.update(visible=False), gr.update(visible=False), gr.update(choices=[], visible=False)

    data = response.json()
    products = data.get("products", [])
    product_cache = []
    dropdown_choices = []

    display_rows = []
    for i, p in enumerate(products):
        if not p.get("product_name"):
            continue
        product_cache.append(p)
        product_display = f"{p.get('product_name', 'N/A')} | {p.get('packaging', 'N/A')} | {p.get('categories', 'N/A')}"
        dropdown_choices.append(product_display)
        display_rows.append([
            p.get("product_name", "N/A"),
            p.get("packaging", "N/A"),
            p.get("categories", "N/A"),
            p.get("code", "")
        ])

    return display_rows, gr.update(visible=True), gr.update(visible=True), gr.update(choices=dropdown_choices, visible=True)

# Step 2: Score product

def score_product(p):
    packaging = (p.get("packaging") or "").lower()
    categories = (p.get("categories") or "").lower()
    ingredients = (p.get("ingredients_text") or "").lower()

    score = 50  # base score

    # Packaging impact
    if "plastic" in packaging:
        score -= 25
    elif "glass" in packaging:
        score += 10
    elif "paper" in packaging or "cardboard" in packaging:
        score += 5
    elif "recyclable" in packaging:
        score += 8
    elif "biodegradable" in packaging or "compostable" in packaging:
        score += 12

    # Category-based adjustments
    if "organic" in categories:
        score += 10
    if "local" in categories:
        score += 5
    if "meat" in categories or "processed food" in categories:
        score -= 15
    if "vegan" in categories:
        score += 10
    if "snacks" in categories or "ready meals" in categories:
        score -= 10

    # Ingredient clues
    if "palm oil" in ingredients:
        score -= 10
    if "organic" in ingredients:
        score += 5
    if "artificial flavor" in ingredients or "preservative" in ingredients:
        score -= 5

    # Normalize score
    return max(0, min(score, 100))
    
# Step 3: Groq explanation

def generate_explanation(p, score):
    prompt = f"""
You are an environmental sustainability expert.

Product:
- Name: {p.get('product_name')}
- Packaging: {p.get('packaging')}
- Categories: {p.get('categories')}
- Ingredients: {p.get('ingredients_text')}
- Sustainability Score: {score}/100

Explain why this product received this score, and suggest 1–2 greener alternatives or improvements.
"""
    try:
        res = client.chat.completions.create(
            model="llama3-8b-8192",
            messages=[{"role": "user", "content": prompt}],
            temperature=0.7
        )
        return res.choices[0].message.content
    except Exception as e:
        return f"Groq explanation failed: {str(e)}"

# Step 4: Analyze selected product

def analyze_selected(product_label):
    global product_cache
    index = -1
    for i, p in enumerate(product_cache):
        full_label = f"{p.get('product_name', 'N/A')} | {p.get('packaging', 'N/A')} | {p.get('categories', 'N/A')}"
        if full_label == product_label:
            index = i
            break

    if index == -1:
        return 0, {}, "Invalid selection", ""

    p = product_cache[index]
    score = score_product(p)
    breakdown = {
        "Product Name": p.get("product_name"),
        "Packaging": p.get("packaging"),
        "Categories": p.get("categories"),
        "Ingredients": p.get("ingredients_text"),
        "Product URL": f"https://world.openfoodfacts.org/product/{p.get('code')}"
    }
    alternatives = suggest_greener_alternatives(p)
    explanation = generate_explanation(p, score)
    return score, breakdown, "\n".join(alternatives), explanation

# Step-5 Suggest Greener Product
def suggest_greener_alternatives(p):
    alternatives = []
    name = p.get("product_name", "this product")
    packaging = (p.get("packaging") or "").lower()
    categories = (p.get("categories") or "").lower()
    ingredients = (p.get("ingredients_text") or "").lower()

    if "plastic" in packaging:
        alternatives.append("Choose items with recyclable, glass, or biodegradable packaging.")
    if "meat" in categories:
        alternatives.append("Consider plant-based or vegan alternatives to reduce emissions.")
    if "palm oil" in ingredients:
        alternatives.append("Look for products without palm oil or with certified sustainable sources.")
    if "ready meal" in categories:
        alternatives.append("Try preparing similar meals at home using local ingredients.")

    if not alternatives:
        alternatives.append(f"{name} is relatively sustainable. You can further reduce impact by buying local or in bulk.")

    return alternatives


# Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("# 🌱 Sustainability Score Analyzer")

    with gr.Row():
        product_input = gr.Textbox(label="Search for a Product")
        search_btn = gr.Button("Search")

    product_table = gr.Dataframe(headers=["Product Name", "Packaging", "Categories", "Code"],
                                 interactive=False, visible=False, label="Matching Products")

    product_selector = gr.Dropdown(label="Select a Product from Search Results", choices=[], visible=False)
    analyze_btn = gr.Button("Analyze Selected Product", visible=False)

    score_output = gr.Number(label="Sustainability Score")
    breakdown_output = gr.JSON(label="Product Breakdown")
    alt_output = gr.Textbox(label="Greener Alternatives")
    explanation_output = gr.Textbox(label="AI Explanation", lines=8)

    search_btn.click(fn=search_products,
                     inputs=[product_input],
                     outputs=[product_table, product_selector, analyze_btn, product_selector])

    analyze_btn.click(fn=analyze_selected,
                      inputs=[product_selector],
                      outputs=[score_output, breakdown_output, alt_output, explanation_output])

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