-def parse(content: bytes) -> BeautifulSoup:
- return BeautifulSoup(content, '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())
-
-