X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/23f3187945d1cf4f3a9cdc43462be2ca39e7023a..8be20b9d36c49271ed392038750dfba981c283b6:/paperdoorknob.py?ds=sidebyside diff --git a/paperdoorknob.py b/paperdoorknob.py index 653d0c6..2bcf31a 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -20,14 +20,6 @@ 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 @@ -46,8 +38,9 @@ def replies(html: BeautifulSoup) -> Iterable[Tag]: def process(spec: Spec) -> None: spec.texout.write(b'\\documentclass{article}\n\\begin{document}\n') - html = clean(parse(spec.fetcher.fetch(spec.url))) + html = parse(spec.htmlfilter(spec.fetcher.fetch(spec.url))) for r in replies(html): + spec.domfilter(r) spec.texout.write(spec.texifier.texify(r)) spec.texout.write(b'\\end{document}\n')