X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/1fac41bf2d403d35d43dd532a37a8225f682bae2..21e82200b2887fc4f4f9eff4023f4a73547cacb5:/glowfic.py diff --git a/glowfic.py b/glowfic.py index 012bd06..92ae4ab 100644 --- a/glowfic.py +++ b/glowfic.py @@ -54,20 +54,31 @@ class Chunk: # * Humans in the community tend to use "posts" to mean chunks. -def chunkDOMs(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()) +class Thread: + + def __init__(self, dom: BeautifulSoup) -> None: + self._dom = dom + + def title(self) -> str | None: + span = self._dom.findChild("span", id="post-title") + if not isinstance(span, Tag): + return None + return span.text.strip() + + def chunkDOMs(self) -> Iterable[Tag]: + def text() -> Tag: + body = self._dom.body + assert body + text = body.find_next("div", class_="post-post") + assert isinstance(text, Tag) + return text + + def the_replies() -> Iterable[Tag]: + rs = self._dom.find_all("div", class_="post-reply") + assert all(isinstance(r, Tag) for r in rs) + return rs + + return itertools.chain([text()], the_replies()) def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk: