X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/b03ad5dc018cc85fd5905011d74de1cc4c234cd9..136277e30143cd1219c896bb4980027ac3c6dbe1:/paperdoorknob.py diff --git a/paperdoorknob.py b/paperdoorknob.py index c684739..bb8bdd1 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -6,11 +6,20 @@ from argparse import ArgumentParser +import os.path +from bs4 import BeautifulSoup import requests +import requests_cache +from xdg_base_dirs import xdg_cache_home def command_line_parser() -> ArgumentParser: parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic') + parser.add_argument( + '--cache_path', + metavar='PATH', + help='Where to keep the http cache (instead of %(default)s)', + default=os.path.join(xdg_cache_home(), "paperdoorknob")) parser.add_argument( '--timeout', help='How long to wait for HTTP requests, in seconds', @@ -19,15 +28,16 @@ def command_line_parser() -> ArgumentParser: return parser -def fetch(url: str, timeout: int) -> None: - with requests.session() as s: - with s.get(url, timeout=timeout) as r: - r.raise_for_status() +def fetch(url: str, session: requests.Session, timeout: int) -> BeautifulSoup: + with session.get(url, timeout=timeout) as r: + r.raise_for_status() + return BeautifulSoup(r.text, 'html.parser') def main() -> None: args = command_line_parser().parse_args() - fetch(args.url, args.timeout) + with requests_cache.CachedSession(args.cache_path, cache_control=True) as session: + fetch(args.url, session, args.timeout) if __name__ == '__main__':