From 23f3187945d1cf4f3a9cdc43462be2ca39e7023a Mon Sep 17 00:00:00 2001 From: Scott Worley Date: Tue, 19 Dec 2023 17:29:14 -0800 Subject: [PATCH 1/1] Bundle the things needed for a run together into a Spec --- args.py | 47 ++++++++++++++++++++++++++++++++++++++ paperdoorknob.py | 53 +++++++++---------------------------------- paperdoorknob_test.py | 29 +++++++++++++++-------- setup.py | 2 ++ spec.py | 21 +++++++++++++++++ 5 files changed, 100 insertions(+), 52 deletions(-) create mode 100644 args.py create mode 100644 spec.py diff --git a/args.py b/args.py new file mode 100644 index 0000000..2a03836 --- /dev/null +++ b/args.py @@ -0,0 +1,47 @@ +# paperdoorknob: Print glowfic +# +# This program is free software: you can redistribute it and/or modify it +# under the terms of the GNU General Public License as published by the +# Free Software Foundation, version 3. + + +from argparse import ArgumentParser +from contextlib import contextmanager +import os.path + +from typing import Iterator + +from xdg_base_dirs import xdg_cache_home + +from fetch import CachingFetcher +from spec import Spec +from texify import PandocTexifier + + +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( + '--out', + help='The filename stem at which to write output ' + + '(eg: "%(default)s" produces %(default)s.tex, %(default)s.pdf, etc.)', + default='book') + parser.add_argument('--pandoc', help='Location of the pandoc executable') + 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 + + +@contextmanager +def spec_from_commandline_args() -> Iterator[Spec]: + args = _command_line_parser().parse_args() + with CachingFetcher(args.cache_path, args.timeout) as fetcher: + with open(args.out + '.tex', 'wb') as texout: + yield Spec(args.url, fetcher, PandocTexifier(args.pandoc or 'pandoc'), texout) diff --git a/paperdoorknob.py b/paperdoorknob.py index 3b86cb8..653d0c6 100644 --- a/paperdoorknob.py +++ b/paperdoorknob.py @@ -5,39 +5,15 @@ # Free Software Foundation, version 3. -from argparse import ArgumentParser import itertools -import os.path -from typing import IO, Iterable +from typing import Iterable from bs4 import BeautifulSoup from bs4.element import Tag -from xdg_base_dirs import xdg_cache_home - -from fetch import CachingFetcher, Fetcher -from texify import PandocTexifier, Texifier - - -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( - '--out', - help='The filename stem at which to write output ' + - '(eg: "%(default)s" produces %(default)s.text, %(default)s.pdf, etc.)', - default='book') - parser.add_argument('--pandoc', help='Location of the pandoc executable') - 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 + +from args import spec_from_commandline_args +from spec import Spec def parse(content: bytes) -> BeautifulSoup: @@ -68,24 +44,17 @@ def replies(html: BeautifulSoup) -> Iterable[Tag]: return itertools.chain([text()], the_replies()) -def process( - url: str, - fetcher: Fetcher, - texifier: Texifier, - texout: IO[bytes]) -> None: - texout.write(b'\\documentclass{article}\n\\begin{document}\n') - html = clean(parse(fetcher.fetch(url))) +def process(spec: Spec) -> None: + spec.texout.write(b'\\documentclass{article}\n\\begin{document}\n') + html = clean(parse(spec.fetcher.fetch(spec.url))) for r in replies(html): - texout.write(texifier.texify(r)) - texout.write(b'\\end{document}\n') + spec.texout.write(spec.texifier.texify(r)) + spec.texout.write(b'\\end{document}\n') def main() -> None: - args = command_line_parser().parse_args() - texifier = PandocTexifier(args.pandoc or 'pandoc') - with CachingFetcher(args.cache_path, args.timeout) as fetcher: - with open(args.out + '.tex', 'wb') as texout: - process(args.url, fetcher, texifier, texout) + with spec_from_commandline_args() as spec: + process(spec) if __name__ == '__main__': diff --git a/paperdoorknob_test.py b/paperdoorknob_test.py index 4cead97..baf7945 100644 --- a/paperdoorknob_test.py +++ b/paperdoorknob_test.py @@ -8,9 +8,12 @@ import unittest import io import subprocess + import paperdoorknob + from testing.fakeserver import FakeGlowficServer from fetch import DirectFetcher +from spec import Spec from texify import DirectTexifier, PandocTexifier, VerifyingTexifier TIMEOUT = 8 @@ -31,11 +34,14 @@ class TestPaperDoorknob(unittest.TestCase): ["This is glowfic", "You sure?", "Pretty sure."]) def testProcess(self) -> None: - texifier = PandocTexifier('pandoc') with DirectFetcher(TIMEOUT) as f: buf = io.BytesIO() - paperdoorknob.process( - f"http://localhost:{self._port}", f, texifier, buf) + spec = Spec( + f"http://localhost:{self._port}", + f, + PandocTexifier('pandoc'), + buf) + paperdoorknob.process(spec) self.assertEqual(buf.getvalue(), b'''\\documentclass{article} \\begin{document} This is glowfic @@ -49,17 +55,20 @@ Pretty sure. PandocTexifier('pandoc'), DirectTexifier()) with DirectFetcher(TIMEOUT) as f: buf = io.BytesIO() - paperdoorknob.process( - f"http://localhost:{self._port}", f, texifier, buf) + spec = Spec(f"http://localhost:{self._port}", f, texifier, buf) + paperdoorknob.process(spec) def testPDF(self) -> None: - texifier = PandocTexifier('pandoc') with DirectFetcher(TIMEOUT) as f: with open("test.tex", 'wb') as out: - paperdoorknob.process( - f"http://localhost:{self._port}", f, texifier, out) - subprocess.run(['pdflatex', 'test.tex'], - stdin=subprocess.DEVNULL, check=True) + spec = Spec( + f"http://localhost:{self._port}", + f, + PandocTexifier('pandoc'), + out) + paperdoorknob.process(spec) + subprocess.run(['pdflatex', 'test.tex'], + stdin=subprocess.DEVNULL, check=True) if __name__ == '__main__': diff --git a/setup.py b/setup.py index 6464879..edaa429 100644 --- a/setup.py +++ b/setup.py @@ -8,8 +8,10 @@ setup( author_email="scottworley@scottworley.com", url="https://git.scottworley.com/paperdoorknob", py_modules=[ + 'args', 'fetch', 'paperdoorknob', + 'spec', 'texify', ], license="GPL-3.0", diff --git a/spec.py b/spec.py new file mode 100644 index 0000000..1322897 --- /dev/null +++ b/spec.py @@ -0,0 +1,21 @@ +# paperdoorknob: Print glowfic +# +# This program is free software: you can redistribute it and/or modify it +# under the terms of the GNU General Public License as published by the +# Free Software Foundation, version 3. + + +from dataclasses import dataclass + +from typing import IO + +from fetch import Fetcher +from texify import Texifier + + +@dataclass(frozen=True) +class Spec: + url: str + fetcher: Fetcher + texifier: Texifier + texout: IO[bytes] -- 2.44.1