X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/47cfa3cdb4b4a7c6191e4a63b2a3080b52f076b0..a2d424680628f747d2ec20f2ae3e3758f0d12cc5:/paperdoorknob.py diff --git a/paperdoorknob.py b/paperdoorknob.py index 8329f83..91909a3 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -9,7 +9,7 @@ from argparse import ArgumentParser import itertools import os.path -from typing import Iterable +from typing import IO, Iterable from bs4 import BeautifulSoup from bs4.element import Tag @@ -25,6 +25,11 @@ def command_line_parser() -> ArgumentParser: metavar='PATH', help='Where to keep the http cache (instead of %(default)s)', default=os.path.join(xdg_cache_home(), "paperdoorknob")) + parser.add_argument( + '--out', + help='The filename stem at which to write output ' + + '(eg: "%(default)s" produces %(default)s.text, %(default)s.pdf, etc.)', + default='book') parser.add_argument( '--timeout', help='How long to wait for HTTP requests, in seconds', @@ -66,15 +71,19 @@ def replies(html: BeautifulSoup) -> Iterable[Tag]: def process( url: str, session: requests.Session, - timeout: int) -> Iterable[Tag]: + timeout: int, + texout: IO[str], +) -> None: html = clean(fetch(url, session, timeout)) - return replies(html) + replies(html) + print("soon", file=texout) def main() -> None: args = command_line_parser().parse_args() with requests_cache.CachedSession(args.cache_path, cache_control=True) as session: - process(args.url, session, args.timeout) + with open(args.out + '.tex', 'w', encoding='UTF-8') as texout: + process(args.url, session, args.timeout, texout) if __name__ == '__main__':