#!/usr/bin/env python3 import warnings warnings.filterwarnings("ignore", category=DeprecationWarning) import logging from mdf import Database, retronews from argparse import ArgumentParser database = Database() if __name__ == '__main__': parser = ArgumentParser() parser.add_argument('--output', type=str, required=True, help='output directory') parser.add_argument('--from-date', type=str) parser.add_argument('--to-date', type=str) parser.add_argument('--merge-threads', default=retronews.TILE_MERGING_POOL_SIZE, type=int) parser.add_argument('--fetch-threads', default=retronews.PAGE_FETCHING_POOL_SIZE, type=int) parser.add_argument('--only-fetch', action='store_true', help='only fetch magazine tiles and exit, do not merge anything') parser.add_argument('--force-overwrite', action='store_true', help='if file yyyy-mm-dd.pdf already exists, delete it and start over') parser.add_argument('--force-probe', action='store_true', help='force all pages to use the \'probe\' method') parser.add_argument('--fetch-probe-pages', nargs='+', type=int, help='force some pages to use the \'probe\' method, when count of vertical and horizontal tiles is unknown') args = parser.parse_args() retronews.set_tile_merging_pool_size(args.merge_threads) retronews.set_page_fetching_pool_size(args.fetch_threads) for doc in database.get_documents((args.from_date, args.to_date)): url = doc['url'] print(f'grabbing {url}...') if not retronews.grab_magazine(url, output_root=args.output, probe_pages=args.fetch_probe_pages, probe_all=args.force_probe, only_fetch=args.only_fetch, force_overwrite=args.force_overwrite): logging.error(f'failed to grab {url}')