X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/8be20b9d36c49271ed392038750dfba981c283b6..131deef1eba442d6e05c2ff6c6f7669b4b9e2b24:/paperdoorknob.py diff --git a/paperdoorknob.py b/paperdoorknob.py index 2bcf31a..b364f4d 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -5,14 +5,10 @@ # Free Software Foundation, version 3. -import itertools - -from typing import Iterable - from bs4 import BeautifulSoup -from bs4.element import Tag from args import spec_from_commandline_args +from glowfic import chunkDOMs, flatURL, makeChunk from spec import Spec @@ -20,28 +16,25 @@ def parse(content: bytes) -> BeautifulSoup: return BeautifulSoup(content, 'html.parser') -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(spec: Spec) -> None: - spec.texout.write(b'\\documentclass{article}\n\\begin{document}\n') - html = parse(spec.htmlfilter(spec.fetcher.fetch(spec.url))) - for r in replies(html): + spec.texout.write(b'''\\documentclass{article} +\\usepackage{booktabs} +\\usepackage{graphicx} +\\usepackage{longtable} +\\usepackage{varwidth} +\\usepackage{wrapstuff} +''') + if spec.geometry is not None: + spec.texout.write(b'\\usepackage[' + + spec.geometry.encode('UTF-8') + + b']{geometry}\n') + spec.texout.write(b'\\begin{document}\n') + url = flatURL(spec.url) + html = parse(spec.htmlfilter(spec.fetcher.fetch(url))) + for r in chunkDOMs(html): spec.domfilter(r) - spec.texout.write(spec.texifier.texify(r)) + chunk = makeChunk(r, spec.images) + spec.texout.write(spec.texfilter(spec.layout.renderChunk(chunk))) spec.texout.write(b'\\end{document}\n')