X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/940270992010ea5b4c912eb02f502923c0487a31..19ed28f37b59e480cebe8f122e07ce18ce536db8:/glowfic.py diff --git a/glowfic.py b/glowfic.py index 372b3aa..5ffaac3 100644 --- a/glowfic.py +++ b/glowfic.py @@ -15,6 +15,7 @@ from bs4 import BeautifulSoup from bs4.element import Tag from images import ImageStore +from spec import Spec from texify import Texifier @@ -56,19 +57,28 @@ class Chunk: class Thread: - def __init__(self, html: BeautifulSoup) -> None: - self._html = html + def __init__(self, spec: Spec) -> None: + spec.log('Fetching HTML...\r') + html = spec.fetcher.fetch(flatURL(spec.url)) + spec.log('Parsing HTML...\r') + self._dom = BeautifulSoup(spec.htmlfilter(html), 'html.parser') + + 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._html.body + 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._html.find_all("div", class_="post-reply") + rs = self._dom.find_all("div", class_="post-reply") assert all(isinstance(r, Tag) for r in rs) return rs