X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/1452f8d33eb6ffe6812698768f1e6bb975d9c4cf..228bb7bb52ef9df820fb41312144c87ee987434d:/paperdoorknob.py diff --git a/paperdoorknob.py b/paperdoorknob.py index 4ee4dbc..72ec1a5 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -4,6 +4,7 @@ # under the terms of the GNU General Public License as published by the # Free Software Foundation, version 3. +from typing import Any, Iterable from bs4 import BeautifulSoup @@ -16,9 +17,16 @@ 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{booktabs} \\usepackage{graphicx} +\\usepackage{longtable} +\\usepackage{soul} \\usepackage{varwidth} \\usepackage{wrapstuff} ''') @@ -26,13 +34,23 @@ def process(spec: Spec) -> None: spec.texout.write(b'\\usepackage[' + spec.geometry.encode('UTF-8') + b']{geometry}\n') - spec.texout.write(b'\\begin{document}\n') + spec.texout.write(b'''\\begin{document} +\\newcommand{\\href}[2]{#2\\footnote{\\detokenize{#1}}} +''') url = flatURL(spec.url) - html = parse(spec.htmlfilter(spec.fetcher.fetch(url))) - for r in 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) chunk = makeChunk(r, spec.images) - spec.texout.write(spec.layout.renderChunk(chunk)) + spec.texout.write(spec.texfilter(spec.layout.renderChunk(chunk))) + spec.log('') spec.texout.write(b'\\end{document}\n')