X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/ba3b7c52144a19a83892638415d2bfdb667888f5..0374fbde6508c909522c701cca9c7ca672a66719:/paperdoorknob.py diff --git a/paperdoorknob.py b/paperdoorknob.py index 3aae538..24ccccd 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -5,37 +5,47 @@ # Free Software Foundation, version 3. -from argparse import ArgumentParser -import os.path -import requests -import requests_cache -from xdg_base_dirs import xdg_cache_home - - -def command_line_parser() -> ArgumentParser: - parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic') - parser.add_argument( - '--cache_path', - metavar='PATH', - help='Where to keep the http cache (instead of %(default)s)', - default=os.path.join(xdg_cache_home(), "paperdoorknob")) - parser.add_argument( - '--timeout', - help='How long to wait for HTTP requests, in seconds', - default=30) - parser.add_argument('url', help='URL to retrieve') - return parser - - -def fetch(url: str, session: requests.Session, timeout: int) -> None: - with session.get(url, timeout=timeout) as r: - r.raise_for_status() +from bs4 import BeautifulSoup + +from args import spec_from_commandline_args +from glowfic import chunkDOMs, makeChunk +from spec import Spec + + +def parse(content: bytes) -> BeautifulSoup: + return BeautifulSoup(content, 'html.parser') + + +def process(spec: Spec) -> None: + spec.texout.write(b'''\\documentclass{article} +\\usepackage{graphicx} +\\usepackage{wrapfig} +''') + 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') + html = parse(spec.htmlfilter(spec.fetcher.fetch(spec.url))) + first = True + for r in chunkDOMs(html): + if first: + first = False + else: + # h/t https://tex.stackexchange.com/questions/309856 + spec.texout.write(b''' +\\vspace{-.5\\ht\\strutbox}\\noindent\\hrulefill + +''') + spec.domfilter(r) + chunk = makeChunk(r, spec.images) + spec.texout.write(spec.layout.renderChunk(chunk)) + spec.texout.write(b'\\end{document}\n') def main() -> None: - args = command_line_parser().parse_args() - with requests_cache.CachedSession(args.cache_path) as session: - fetch(args.url, session, args.timeout) + with spec_from_commandline_args() as spec: + process(spec) if __name__ == '__main__':