Files
mobile.de/iaai_scraper/cli.py
2026-04-08 18:30:02 +03:00

134 lines
6.2 KiB
Python

import argparse
from pathlib import Path
from .core.config import Settings
from .core.utils import save_to_json
from .scraper import IAAIScraper
def build_parser() -> argparse.ArgumentParser:
# Описание CLI-команд и их аргументов.
parser = argparse.ArgumentParser(description="Public IAAI scraper")
parser.add_argument("--headless", choices=["true", "false"], default=None, help="Override headless mode")
parser.add_argument("--debug", action="store_true", help="Enable DEBUG logging")
subparsers = parser.add_subparsers(dest="command", required=True)
default_output_dir = Path("artifacts/json")
init_db_parser = subparsers.add_parser("init-db", help="Create local DB tables")
init_db_parser.add_argument("--output", default=str(default_output_dir / "iaai_db_init.json"), help="Path to output JSON")
listing_parser = subparsers.add_parser("collect-listing", help="Collect vehicle URLs from Vehiclelisting/Cars")
listing_parser.add_argument("--make", default=None, help="Optional make filter")
listing_parser.add_argument("--model", default=None, help="Optional model filter")
listing_parser.add_argument("--output", default=str(default_output_dir / "iaai_listing_links.json"), help="Path to output JSON")
open_parser = subparsers.add_parser(
"open-vehicle",
help="Open a vehicle page in gentle mode and save only DOM-based hints",
)
open_parser.add_argument("vehicle_url", help="IAAI vehicle detail URL")
open_parser.add_argument("--output", default=str(default_output_dir / "iaai_vehicle_opened.json"), help="Path to output JSON")
scrape_parser = subparsers.add_parser(
"scrape-vehicle",
help="Open a vehicle page and capture a limited set of likely useful JSON responses",
)
scrape_parser.add_argument("vehicle_url", help="IAAI vehicle detail URL")
scrape_parser.add_argument("--output", default=str(default_output_dir / "iaai_vehicle_detail.json"), help="Path to output JSON")
export_parser = subparsers.add_parser(
"export-db-json",
help="Scrape a vehicle page and save only the DB-ready car record JSON",
)
export_parser.add_argument("vehicle_url", help="IAAI vehicle detail URL")
export_parser.add_argument("--output", default=str(default_output_dir / "iaai_vehicle_db_record.json"), help="Path to output JSON")
sync_vehicle_parser = subparsers.add_parser("sync-vehicle", help="Scrape one vehicle and upsert it into the DB")
sync_vehicle_parser.add_argument("vehicle_url", help="IAAI vehicle detail URL")
sync_vehicle_parser.add_argument("--lane", default="iaai", help="Logical lane name for sync_runs")
sync_vehicle_parser.add_argument("--output", default=str(default_output_dir / "iaai_sync_vehicle.json"), help="Path to output JSON")
sync_listing_parser = subparsers.add_parser(
"sync-listing",
help="Collect vehicle URLs from the Cars listing and sync them sequentially",
)
sync_listing_parser.add_argument("--make", default=None, help="Optional make filter")
sync_listing_parser.add_argument("--model", default=None, help="Optional model filter")
sync_listing_parser.add_argument("--lane", default="iaai_cars", help="Logical lane name for sync_runs")
sync_listing_parser.add_argument("--limit", type=int, default=None, help="Limit number of vehicles to sync")
sync_listing_parser.add_argument(
"--only-new",
choices=["true", "false"],
default=None,
help="Process only new vehicles (default from IAAI_SYNC_ONLY_NEW)",
)
sync_listing_parser.add_argument("--output", default=str(default_output_dir / "iaai_sync_listing.json"), help="Path to output JSON")
daemon_parser = subparsers.add_parser(
"run-daemon",
help="Run the scraper in a loop, syncing vehicles every N minutes (default 60)",
)
daemon_parser.add_argument(
"--interval", type=int, default=None,
help="Override interval in minutes (default from IAAI_SCHEDULER_INTERVAL_MINUTES or 60)",
)
return parser
def main() -> None:
# Собираем runtime overrides только при необходимости.
parser = build_parser()
args = parser.parse_args()
runtime_settings: Settings | None = None
if args.headless is not None or args.debug:
runtime_settings = Settings()
if args.headless is not None:
runtime_settings.headless = args.headless == "true"
if args.debug:
runtime_settings.log_level = "DEBUG"
if args.command == "run-daemon":
# daemon: бесконечный цикл
runtime_settings = runtime_settings or Settings()
if args.interval is not None:
runtime_settings.scheduler_interval_minutes = args.interval
with IAAIScraper(runtime_settings) as scraper:
scraper.persistence.create_tables()
scraper.run_scheduled()
return
# одноразовый запуск
with IAAIScraper(runtime_settings) as scraper:
# Каждая команда сводится к одному методу скрапера.
if args.command == "init-db":
data = scraper.init_db()
elif args.command == "collect-listing":
data = scraper.collect_listing(make=args.make, model=args.model)
elif args.command == "open-vehicle":
data = scraper.open_vehicle_page(args.vehicle_url)
elif args.command == "scrape-vehicle":
data = scraper.scrape_vehicle_detail(args.vehicle_url)
elif args.command == "export-db-json":
data = scraper.scrape_vehicle_detail(args.vehicle_url).get("db_record", {})
elif args.command == "sync-vehicle":
data = scraper.sync_vehicle(args.vehicle_url, lane=args.lane)
else:
only_new = None if args.only_new is None else args.only_new == "true"
data = scraper.sync_listing(
make=args.make,
model=args.model,
lane=args.lane,
limit=args.limit,
only_new=only_new,
)
save_to_json(data, Path(args.output))
print(f"Saved result to {Path(args.output).resolve()}")
if __name__ == "__main__":
main()