X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/b254976424d0f56fab14f8897dade345ce46a2d1..7804859f8fe8b8dc94b6a33b2221ac0b68d7ed86:/paperdoorknob.py diff --git a/paperdoorknob.py b/paperdoorknob.py index 6a55675..a19045c 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -22,25 +22,29 @@ def ilen(it: Iterable[Any]) -> int: def process(spec: Spec) -> None: - spec.texout.write(b'''\\documentclass{article} -\\usepackage{booktabs} -\\usepackage{graphicx} -\\usepackage{longtable} -\\usepackage{soul} -\\usepackage{varwidth} -\\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} -\\newcommand{\\href}[2]{#2\\footnote{#1}} + spec.texout.write(br'''\begin{document} +\newcommand{\href}[2]{#2\footnote{\detokenize{#1}}} ''') url = flatURL(spec.url) - html = parse(spec.htmlfilter(spec.fetcher.fetch(url))) - num_chunks = ilen(chunkDOMs(html)) - for i, r in enumerate(chunkDOMs(html)): + spec.log('Fetching HTML...\r') + html = spec.fetcher.fetch(url) + spec.log('Parsing HTML...\r') + dom = parse(spec.htmlfilter(html)) + spec.log('Counting chunks...\r') + num_chunks = ilen(chunkDOMs(dom)) + for i, r in enumerate(chunkDOMs(dom)): percent = 100.0 * i / num_chunks spec.log(f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)\r') spec.domfilter(r)