X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/5f230208bb5e858cfeadb3d9c822579d2b7b8300..940270992010ea5b4c912eb02f502923c0487a31:/paperdoorknob.py?ds=sidebyside diff --git a/paperdoorknob.py b/paperdoorknob.py index 0fb3fe1..ade0b88 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -7,9 +7,10 @@ from typing import Any, Iterable from bs4 import BeautifulSoup +from bs4.element import Tag from args import spec_from_commandline_args -from glowfic import chunkDOMs, flatURL, makeChunk +from glowfic import flatURL, makeChunk, renderChunk, Thread from spec import Spec @@ -21,6 +22,13 @@ def ilen(it: Iterable[Any]) -> int: return sum(1 for _ in it) +def get_title(dom: BeautifulSoup) -> str | None: + span = dom.findChild("span", id="post-title") + if not isinstance(span, Tag): + return None + return span.text + + def process(spec: Spec) -> None: spec.texout.write(br'''\documentclass{article} \usepackage{booktabs} @@ -40,20 +48,25 @@ def process(spec: Spec) -> None: \newcommand{\glowicon}[1]{\includegraphics[ width=\glowiconsize,height=\glowiconsize,keepaspectratio ]{#1}} -''' % spec.icon_size) +\newcommand{\ifnotempty}[2]{\expandafter\ifx\expandafter\relax + \detokenize{#1}\relax\else #2\fi} +%s +''' % (spec.icon_size, spec.layout)) url = flatURL(spec.url) spec.log('Fetching HTML...\r') html = spec.fetcher.fetch(url) spec.log('Parsing HTML...\r') dom = parse(spec.htmlfilter(html)) + thread = Thread(dom) spec.log('Counting chunks...\r') - num_chunks = ilen(chunkDOMs(dom)) - for i, r in enumerate(chunkDOMs(dom)): + num_chunks = ilen(thread.chunkDOMs()) + title = get_title(dom) or "chunk" + for i, r in enumerate(thread.chunkDOMs()): percent = 100.0 * i / num_chunks - spec.log(f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)\r') + spec.log(f'Processing {title} {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.texout.write(spec.texfilter(renderChunk(spec.texifier, chunk))) spec.log('') spec.texout.write(b'\\end{document}\n')