add mobilede scraper
This commit is contained in:
127
mobilede_scraper/cli.py
Normal file
127
mobilede_scraper/cli.py
Normal file
@@ -0,0 +1,127 @@
|
||||
import argparse
|
||||
from pathlib import Path
|
||||
|
||||
from .core.config import Settings
|
||||
from .core.utils import save_to_json
|
||||
from .mobile_de import MobileDeClient, MobileDeScraper
|
||||
|
||||
|
||||
def build_parser() -> argparse.ArgumentParser:
|
||||
parser = argparse.ArgumentParser(description="mobile.de scraper CLI")
|
||||
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")
|
||||
|
||||
subparsers.add_parser("init-db", help="Create DB tables")
|
||||
|
||||
mobile_search_parser = subparsers.add_parser("search", aliases=["mobilede-search"], help="Collect mobile.de search result pages")
|
||||
mobile_search_parser.add_argument("--page", type=int, default=1)
|
||||
mobile_search_parser.add_argument("--max-pages", type=int, default=1)
|
||||
mobile_search_parser.add_argument("--delay", type=float, default=0.7)
|
||||
mobile_search_parser.add_argument("--search-url", default=None, help="готовая mobile.de ссылка с фильтрами")
|
||||
mobile_search_parser.add_argument("--make-id", default=None, help="mobile.de make id, for example BMW=3500")
|
||||
mobile_search_parser.add_argument("--model-id", default=None, help="mobile.de model id")
|
||||
mobile_search_parser.add_argument("--price-min", default=None)
|
||||
mobile_search_parser.add_argument("--price-max", default=None)
|
||||
mobile_search_parser.add_argument("--year-min", default=None)
|
||||
mobile_search_parser.add_argument("--year-max", default=None)
|
||||
mobile_search_parser.add_argument("--output", default=str(default_output_dir / "mobilede_listing_links.json"))
|
||||
|
||||
mobile_detail_parser = subparsers.add_parser("detail", aliases=["mobilede-detail"], help="Collect one mobile.de detail page")
|
||||
mobile_detail_parser.add_argument("listing_id")
|
||||
mobile_detail_parser.add_argument("--output", default=str(default_output_dir / "mobilede_vehicle_detail.json"))
|
||||
|
||||
mobile_sync_search_parser = subparsers.add_parser("sync-search", help="Collect and upsert mobile.de search result pages")
|
||||
mobile_sync_search_parser.add_argument("--page", type=int, default=1)
|
||||
mobile_sync_search_parser.add_argument("--max-pages", type=int, default=1)
|
||||
mobile_sync_search_parser.add_argument("--delay", type=float, default=0.7)
|
||||
mobile_sync_search_parser.add_argument("--lane", default="mobile_de_cars")
|
||||
mobile_sync_search_parser.add_argument("--search-url", default=None, help="готовая mobile.de ссылка с фильтрами")
|
||||
mobile_sync_search_parser.add_argument("--make-id", default=None)
|
||||
mobile_sync_search_parser.add_argument("--model-id", default=None)
|
||||
mobile_sync_search_parser.add_argument("--price-min", default=None)
|
||||
mobile_sync_search_parser.add_argument("--price-max", default=None)
|
||||
mobile_sync_search_parser.add_argument("--year-min", default=None)
|
||||
mobile_sync_search_parser.add_argument("--year-max", default=None)
|
||||
mobile_sync_search_parser.add_argument("--output", default=str(default_output_dir / "mobilede_sync_search.json"))
|
||||
|
||||
mobile_sync_detail_parser = subparsers.add_parser("sync-detail", help="Collect and upsert one mobile.de detail page")
|
||||
mobile_sync_detail_parser.add_argument("listing_id")
|
||||
mobile_sync_detail_parser.add_argument("--lane", default="mobile_de_cars")
|
||||
mobile_sync_detail_parser.add_argument("--output", default=str(default_output_dir / "mobilede_sync_detail.json"))
|
||||
|
||||
return parser
|
||||
|
||||
|
||||
def main() -> None:
|
||||
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 in {"search", "mobilede-search"}:
|
||||
scraper = MobileDeScraper(MobileDeClient(delay_seconds=args.delay))
|
||||
data = scraper.collect_search(
|
||||
start_page=args.page,
|
||||
max_pages=args.max_pages,
|
||||
search_url=args.search_url,
|
||||
make_id=args.make_id,
|
||||
model_id=args.model_id,
|
||||
price_min=args.price_min,
|
||||
price_max=args.price_max,
|
||||
year_min=args.year_min,
|
||||
year_max=args.year_max,
|
||||
)
|
||||
save_to_json(data, Path(args.output))
|
||||
print(f"Saved to {Path(args.output).resolve()}")
|
||||
return
|
||||
|
||||
if args.command in {"detail", "mobilede-detail"}:
|
||||
scraper = MobileDeScraper()
|
||||
data = scraper.collect_detail(args.listing_id)
|
||||
save_to_json(data, Path(args.output))
|
||||
print(f"Saved to {Path(args.output).resolve()}")
|
||||
return
|
||||
|
||||
if args.command == "sync-search":
|
||||
scraper = MobileDeScraper(MobileDeClient(delay_seconds=args.delay))
|
||||
data = scraper.sync_search(
|
||||
start_page=args.page,
|
||||
max_pages=args.max_pages,
|
||||
lane=args.lane,
|
||||
search_url=args.search_url,
|
||||
make_id=args.make_id,
|
||||
model_id=args.model_id,
|
||||
price_min=args.price_min,
|
||||
price_max=args.price_max,
|
||||
year_min=args.year_min,
|
||||
year_max=args.year_max,
|
||||
)
|
||||
save_to_json(data, Path(args.output))
|
||||
print(f"Saved to {Path(args.output).resolve()}")
|
||||
return
|
||||
|
||||
if args.command == "sync-detail":
|
||||
scraper = MobileDeScraper()
|
||||
data = scraper.sync_detail(args.listing_id, lane=args.lane)
|
||||
save_to_json(data, Path(args.output))
|
||||
print(f"Saved to {Path(args.output).resolve()}")
|
||||
return
|
||||
|
||||
if args.command == "init-db":
|
||||
scraper = MobileDeScraper()
|
||||
data = scraper.init_db()
|
||||
print(f"DB initialized: {data}")
|
||||
return
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Reference in New Issue
Block a user