X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/357f37be787409c5655df646f8aa0ed4abfa5eaf..70adfbff16123a62d6b25f8ead23789d831d1d70:/paperdoorknob.py diff --git a/paperdoorknob.py b/paperdoorknob.py index f2a81d2..d545f61 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -4,42 +4,51 @@ # under the terms of the GNU General Public License as published by the # Free Software Foundation, version 3. - -from bs4 import BeautifulSoup +from typing import Any, Iterable from args import spec_from_commandline_args -from glowfic import chunkDOMs, makeChunk +from glowfic import makeChunk, renderChunk, Thread from spec import Spec -def parse(content: bytes) -> BeautifulSoup: - return BeautifulSoup(content, 'html.parser') +def ilen(it: Iterable[Any]) -> int: + return sum(1 for _ in it) def process(spec: Spec) -> None: - spec.texout.write(b'''\\documentclass{article} -\\usepackage{graphicx} -\\usepackage{wrapstuff} + spec.texout.write(br'''\documentclass{article} +\usepackage{booktabs} +\usepackage{graphicx} +\usepackage{longtable} +\usepackage{soul} +\usepackage{varwidth} +\usepackage{wrapstuff} ''') if spec.geometry is not None: - spec.texout.write(b'\\usepackage[' + + spec.texout.write(br'\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''' -\\wrapstuffclear\\vspace{-.5\\ht\\strutbox}\\noindent\\hrulefill - -''') + spec.texout.write(br'''\begin{document} +\newcommand{\href}[2]{#2\footnote{\detokenize{#1}}} +\def\glowiconsize{%fmm} +\newcommand{\glowicon}[1]{\includegraphics[ + width=\glowiconsize,height=\glowiconsize,keepaspectratio +]{#1}} +\newcommand{\ifnotempty}[2]{\expandafter\ifx\expandafter\relax + \detokenize{#1}\relax\else #2\fi} +%s +''' % (spec.icon_size, spec.layout)) + thread = Thread(spec) + spec.log('Counting chunks...\r') + num_chunks = ilen(thread.chunkDOMs()) + title = thread.title() or "chunk" + for i, r in enumerate(thread.chunkDOMs()): + percent = 100.0 * i / num_chunks + spec.log(f'Processing {title} {i} of {num_chunks} ({percent:.1f}%)\r') spec.domfilter(r) chunk = makeChunk(r, spec.images) - spec.texout.write(spec.layout.renderChunk(chunk)) + spec.texout.write(spec.texfilter(renderChunk(spec.texifier, chunk))) + spec.log('') spec.texout.write(b'\\end{document}\n')