]>
Commit | Line | Data |
---|---|---|
1 | # paperdoorknob: Print glowfic | |
2 | # | |
3 | # This program is free software: you can redistribute it and/or modify it | |
4 | # under the terms of the GNU General Public License as published by the | |
5 | # Free Software Foundation, version 3. | |
6 | ||
7 | ||
8 | from argparse import ArgumentParser | |
9 | import itertools | |
10 | import os.path | |
11 | ||
12 | from typing import Iterable | |
13 | ||
14 | from bs4 import BeautifulSoup | |
15 | from bs4.element import Tag | |
16 | import requests | |
17 | import requests_cache | |
18 | from xdg_base_dirs import xdg_cache_home | |
19 | ||
20 | ||
21 | class Post: | |
22 | def __init__(self, html: BeautifulSoup) -> None: | |
23 | self._html = html | |
24 | ||
25 | def text(self) -> Tag: | |
26 | body = self._html.body | |
27 | assert body | |
28 | text = body.find_next("div", class_="post-post") | |
29 | assert isinstance(text, Tag) | |
30 | return text | |
31 | ||
32 | def replies(self) -> Iterable[Tag]: | |
33 | replies = self._html.find_all("div", class_="post-reply") | |
34 | assert all(isinstance(r, Tag) for r in replies) | |
35 | return replies | |
36 | ||
37 | def entries(self) -> Iterable[Tag]: | |
38 | return itertools.chain([self.text()], self.replies()) | |
39 | ||
40 | ||
41 | def command_line_parser() -> ArgumentParser: | |
42 | parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic') | |
43 | parser.add_argument( | |
44 | '--cache_path', | |
45 | metavar='PATH', | |
46 | help='Where to keep the http cache (instead of %(default)s)', | |
47 | default=os.path.join(xdg_cache_home(), "paperdoorknob")) | |
48 | parser.add_argument( | |
49 | '--timeout', | |
50 | help='How long to wait for HTTP requests, in seconds', | |
51 | default=30) | |
52 | parser.add_argument('url', help='URL to retrieve') | |
53 | return parser | |
54 | ||
55 | ||
56 | def fetch(url: str, session: requests.Session, timeout: int) -> BeautifulSoup: | |
57 | with session.get(url, timeout=timeout) as r: | |
58 | r.raise_for_status() | |
59 | return BeautifulSoup(r.text, 'html.parser') | |
60 | ||
61 | ||
62 | def main() -> None: | |
63 | args = command_line_parser().parse_args() | |
64 | with requests_cache.CachedSession(args.cache_path, cache_control=True) as session: | |
65 | html = fetch(args.url, session, args.timeout) | |
66 | Post(html) | |
67 | ||
68 | ||
69 | if __name__ == '__main__': | |
70 | main() |