X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/170e50a0efa18ed0f5f2ff644793469894983029..47cfa3cdb4b4a7c6191e4a63b2a3080b52f076b0:/paperdoorknob.py?ds=inline diff --git a/paperdoorknob.py b/paperdoorknob.py index fb2df1f..8329f83 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -18,30 +18,6 @@ import requests_cache from xdg_base_dirs import xdg_cache_home -class Post: - def __init__(self, html: BeautifulSoup) -> None: - self._html = html - for eb in self._html.find_all("div", class_="post-edit-box"): - eb.decompose() - for footer in self._html.find_all("div", class_="post-footer"): - footer.decompose() - - def text(self) -> Tag: - body = self._html.body - assert body - text = body.find_next("div", class_="post-post") - assert isinstance(text, Tag) - return text - - def replies(self) -> Iterable[Tag]: - replies = self._html.find_all("div", class_="post-reply") - assert all(isinstance(r, Tag) for r in replies) - return replies - - def entries(self) -> Iterable[Tag]: - return itertools.chain([self.text()], self.replies()) - - def command_line_parser() -> ArgumentParser: parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic') parser.add_argument( @@ -63,11 +39,42 @@ def fetch(url: str, session: requests.Session, timeout: int) -> BeautifulSoup: return BeautifulSoup(r.text, 'html.parser') +def clean(html: BeautifulSoup) -> BeautifulSoup: + for eb in html.find_all("div", class_="post-edit-box"): + eb.decompose() + for footer in html.find_all("div", class_="post-footer"): + footer.decompose() + return html + + +def replies(html: BeautifulSoup) -> Iterable[Tag]: + def text() -> Tag: + body = html.body + assert body + text = body.find_next("div", class_="post-post") + assert isinstance(text, Tag) + return text + + def the_replies() -> Iterable[Tag]: + rs = html.find_all("div", class_="post-reply") + assert all(isinstance(r, Tag) for r in rs) + return rs + + return itertools.chain([text()], the_replies()) + + +def process( + url: str, + session: requests.Session, + timeout: int) -> Iterable[Tag]: + html = clean(fetch(url, session, timeout)) + return replies(html) + + def main() -> None: args = command_line_parser().parse_args() with requests_cache.CachedSession(args.cache_path, cache_control=True) as session: - html = fetch(args.url, session, args.timeout) - Post(html) + process(args.url, session, args.timeout) if __name__ == '__main__':