main.py

#!/usr/bin/env python3
"""
Vinted Scraper - CLI entry point.

Scrapes marketplace listings from Vinted using ScrapingAnt API.
"""

import argparse
import os
import sys
from datetime import datetime

from config import (
    CATEGORIES,
    DEFAULT_DELAY,
    DEFAULT_PAGES,
    OUTPUT_DIR,
    DEFAULT_OUTPUT_FILE
)
from scraper import VintedScraper


def parse_args():
    """Parse command line arguments."""
    parser = argparse.ArgumentParser(
        description="Scrape Vinted marketplace listings",
        formatter_class=argparse.RawDescriptionHelpFormatter,
        epilog="""
Examples:
  python main.py -s "nike shoes"           Search for nike shoes
  python main.py -c women -p 3             Scrape women's category, 3 pages
  python main.py -s "vintage" --json       Export to JSON
  python main.py --list-categories         List available categories
        """
    )

    parser.add_argument(
        "-s", "--search",
        type=str,
        default="",
        help="Search query (default: empty)"
    )

    parser.add_argument(
        "-c", "--category",
        type=str,
        default="",
        help="Category to scrape (see --list-categories)"
    )

    parser.add_argument(
        "-p", "--pages",
        type=int,
        default=DEFAULT_PAGES,
        help=f"Maximum pages to scrape (default: {DEFAULT_PAGES})"
    )

    parser.add_argument(
        "-d", "--delay",
        type=float,
        default=DEFAULT_DELAY,
        help=f"Delay between requests in seconds (default: {DEFAULT_DELAY})"
    )

    parser.add_argument(
        "-o", "--output",
        type=str,
        default="",
        help="Output filename (without extension)"
    )

    parser.add_argument(
        "--json",
        action="store_true",
        help="Also export to JSON format"
    )

    parser.add_argument(
        "--min-price",
        type=float,
        default=None,
        help="Minimum price filter"
    )

    parser.add_argument(
        "--max-price",
        type=float,
        default=None,
        help="Maximum price filter"
    )

    parser.add_argument(
        "--list-categories",
        action="store_true",
        help="List available categories and exit"
    )

    parser.add_argument(
        "--api-key",
        type=str,
        default=os.environ.get("SCRAPINGANT_API_KEY", ""),
        help="ScrapingAnt API key (or use SCRAPINGANT_API_KEY env var)"
    )

    return parser.parse_args()


def list_categories():
    """Print available categories."""
    print("\nAvailable categories:")
    print("-" * 40)
    for key, info in CATEGORIES.items():
        print(f"  {key:15} - {info['name']}")
    print()


def main():
    """Main entry point."""
    args = parse_args()

    # Handle list categories
    if args.list_categories:
        list_categories()
        return 0

    # Validate API key
    if not args.api_key:
        print("Error: ScrapingAnt API key is required.")
        print("Set SCRAPINGANT_API_KEY environment variable or use --api-key")
        return 1

    # Validate category if provided
    if args.category and args.category not in CATEGORIES:
        print(f"Error: Invalid category '{args.category}'")
        list_categories()
        return 1

    # Validate that at least search or category is provided
    if not args.search and not args.category:
        print("Error: Please provide either --search or --category")
        print("Use --help for usage information")
        return 1

    # Create output directory
    os.makedirs(OUTPUT_DIR, exist_ok=True)

    # Initialize scraper
    scraper = VintedScraper(
        api_key=args.api_key,
        delay=args.delay
    )

    print(f"\nVinted Scraper")
    print("=" * 50)

    if args.search:
        print(f"Search query: {args.search}")
    if args.category:
        print(f"Category: {args.category}")
    print(f"Max pages: {args.pages}")
    if args.min_price:
        print(f"Min price: ${args.min_price}")
    if args.max_price:
        print(f"Max price: ${args.max_price}")
    print("=" * 50)

    # Scrape listings
    total = scraper.scrape_search(
        search_text=args.search,
        category=args.category,
        max_pages=args.pages,
        min_price=args.min_price,
        max_price=args.max_price
    )

    print(f"\n{'=' * 50}")
    print(f"Total listings scraped: {total}")

    if total == 0:
        print("No listings found.")
        return 0

    # Generate output filename
    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
    if args.output:
        base_name = args.output
    else:
        search_part = args.search.replace(" ", "_")[:20] if args.search else ""
        category_part = args.category if args.category else ""
        base_name = f"{DEFAULT_OUTPUT_FILE}_{search_part or category_part}_{timestamp}"

    # Export results
    csv_path = os.path.join(OUTPUT_DIR, f"{base_name}.csv")
    scraper.export_csv(csv_path)
    print(f"Results exported to: {csv_path}")

    if args.json:
        json_path = os.path.join(OUTPUT_DIR, f"{base_name}.json")
        scraper.export_json(json_path)
        print(f"JSON exported to: {json_path}")

    return 0


if __name__ == "__main__":
    sys.exit(main())