scraper.py

# scraper.py
import os
from apify_client import ApifyClient

APIFY_TOKEN = os.getenv("APIFY_TOKEN")
if not APIFY_TOKEN:
    raise ValueError("APIFY_TOKEN environment variable is required")

ACTOR_ID = os.getenv("APIFY_ACTOR_ID")
if not ACTOR_ID:
    raise ValueError("APIFY_ACTOR_ID environment variable is required")

_client = ApifyClient(APIFY_TOKEN)

def scrape_product(url: str, max_reviews: int = 100):
    """Fetch product details + reviews with your working Apify actor."""
    run_input = {
        "productUrls": [{"url": url}],
        "maxReviews": int(max_reviews),
        "scrapeProductDetails": True,
        "sort": "helpful",
        "filterByRatings": ["allStars"],
    }

    run = _client.actor(ACTOR_ID).call(run_input=run_input)

    product, reviews = None, []
    for item in _client.dataset(run["defaultDatasetId"]).iterate_items():
        # product details
        if not product and ("productTitle" in item or "productAsin" in item):
            product = {
                "asin": item.get("productAsin"),
                "title": item.get("productTitle") or "N/A",
                "price": item.get("productPrice") or "N/A",
                "stars": item.get("productStars") or "N/A",
                "image": item.get("productMainImageUrl"),
                "url": url,
            }

        # reviews
        if item.get("reviewDescription"):
            reviews.append({
                "asin": item.get("productAsin"),
                "rating": item.get("ratingScore"),
                "title": item.get("reviewTitle"),
                "review": item.get("reviewDescription"),
                "date": item.get("date"),
                "country": item.get("country"),
                "verified": item.get("isVerified"),
                "url": item.get("reviewUrl"),
            })

    # Fallback product shell if Apify didn’t return details
    if not product:
        product = {"asin": None, "title": "N/A", "price": "N/A", "stars": "N/A", "image": None, "url": url}

    return product, reviews