-
-from argparse import ArgumentParser
-import itertools
-import os.path
-
-from typing import Iterable
-
-from bs4 import BeautifulSoup
-from bs4.element import Tag
-import requests
-import requests_cache
-from xdg_base_dirs import xdg_cache_home
-
-
-class Post:
- def __init__(self, html: BeautifulSoup) -> None:
- self._html = html
- for eb in self._html.find_all("div", class_="post-edit-box"):
- eb.decompose()
- for footer in self._html.find_all("div", class_="post-footer"):
- footer.decompose()
-
- def text(self) -> Tag:
- body = self._html.body
- assert body
- text = body.find_next("div", class_="post-post")
- assert isinstance(text, Tag)
- return text
-
- def replies(self) -> Iterable[Tag]:
- replies = self._html.find_all("div", class_="post-reply")
- assert all(isinstance(r, Tag) for r in replies)
- return replies
-
- def entries(self) -> Iterable[Tag]:
- return itertools.chain([self.text()], self.replies())
-
-
-def command_line_parser() -> ArgumentParser:
- parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
- parser.add_argument(
- '--cache_path',
- metavar='PATH',
- help='Where to keep the http cache (instead of %(default)s)',
- default=os.path.join(xdg_cache_home(), "paperdoorknob"))
- parser.add_argument(
- '--timeout',
- help='How long to wait for HTTP requests, in seconds',
- default=30)
- parser.add_argument('url', help='URL to retrieve')
- return parser
-
-
-def fetch(url: str, session: requests.Session, timeout: int) -> BeautifulSoup:
- with session.get(url, timeout=timeout) as r:
- r.raise_for_status()
- return BeautifulSoup(r.text, 'html.parser')
+from typing import Any, Iterable
+
+from args import spec_from_commandline_args
+from glowfic import makeChunk, renderChunk, Thread
+from spec import Spec
+
+
+def ilen(it: Iterable[Any]) -> int:
+ return sum(1 for _ in it)
+
+
+def process(spec: Spec) -> None:
+ 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(br'\usepackage[' +
+ spec.geometry.encode('UTF-8') +
+ b']{geometry}\n')
+ 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.texfilter(renderChunk(spec.texifier, chunk)))
+ spec.log('')
+ spec.texout.write(b'\\end{document}\n')