X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/24e866347ec2355c35e0e39e317946b9b833d2ed..372211762c57033dfef33075bd593ec5d426f3c7:/paperdoorknob.py?ds=sidebyside diff --git a/paperdoorknob.py b/paperdoorknob.py index 0da6d97..4e7cdf6 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 @@ -16,6 +17,10 @@ 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} @@ -32,10 +37,14 @@ def process(spec: Spec) -> None: spec.texout.write(b'\\begin{document}\n') url = flatURL(spec.url) html = parse(spec.htmlfilter(spec.fetcher.fetch(url))) - for r in chunkDOMs(html): + num_chunks = ilen(chunkDOMs(html)) + for i, r in enumerate(chunkDOMs(html)): + 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.texfilter(spec.layout.renderChunk(chunk))) + spec.log('') spec.texout.write(b'\\end{document}\n')