from bs4 import BeautifulSoup
from bs4.element import Tag
-import requests
-import requests_cache
from xdg_base_dirs import xdg_cache_home
+from fetch import CachingFetcher, Fetcher
+
def command_line_parser() -> ArgumentParser:
parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
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')
+def fetch(url: str, fetcher: Fetcher) -> BeautifulSoup:
+ return BeautifulSoup(fetcher.fetch(url), 'html.parser')
def clean(html: BeautifulSoup) -> BeautifulSoup:
def process(
url: str,
- session: requests.Session,
- timeout: int,
+ fetcher: Fetcher,
texout: IO[bytes],
- pandoc: str,
-) -> None:
+ pandoc: str) -> None:
texout.write(b'\\documentclass{article}\n\\begin{document}\n')
- html = clean(fetch(url, session, timeout))
+ html = clean(fetch(url, fetcher))
for r in replies(html):
texout.write(html_to_tex(pandoc, r))
texout.write(b'\\end{document}\n')
def main() -> None:
args = command_line_parser().parse_args()
- with requests_cache.CachedSession(args.cache_path, cache_control=True) as session:
+ with CachingFetcher(args.cache_path, args.timeout) as fetcher:
with open(args.out + '.tex', 'wb') as texout:
- process(
- args.url,
- session,
- args.timeout,
- texout,
- args.pandoc or 'pandoc')
+ process(args.url, fetcher, texout, args.pandoc or 'pandoc')
if __name__ == '__main__':
import unittest
import io
import subprocess
-import requests
-import requests_cache
import paperdoorknob
from testing.fakeserver import FakeGlowficServer
+from fetch import DirectFetcher
TIMEOUT = 8
-class TestFetch(unittest.TestCase):
+class TestPaperDoorknob(unittest.TestCase):
def setUp(self) -> None:
self._server = self.enterContext(FakeGlowficServer())
self._port = self._server.port()
- def testFetch(self) -> None:
- with requests.session() as s:
- paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
- self.assertEqual(self._server.request_count(), 1)
- paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
- self.assertEqual(self._server.request_count(), 2)
-
- def testFetchCaching(self) -> None:
- with requests_cache.CachedSession() as s:
- paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
- self.assertEqual(self._server.request_count(), 1)
- paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
- self.assertEqual(self._server.request_count(), 1)
-
- def testFetchPersistentCaching(self) -> None:
- with requests_cache.CachedSession() as s:
- paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
- self.assertEqual(self._server.request_count(), 1)
- with requests_cache.CachedSession() as s:
- paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
- self.assertEqual(self._server.request_count(), 1)
-
def testReplies(self) -> None:
- with requests.session() as s:
+ with DirectFetcher(TIMEOUT) as f:
replies = paperdoorknob.replies(
paperdoorknob.clean(
- paperdoorknob.fetch(
- f"http://localhost:{self._port}",
- s,
- TIMEOUT)))
+ paperdoorknob.fetch(f"http://localhost:{self._port}", f)))
self.assertEqual([r.text.strip() for r in replies],
["This is glowfic", "You sure?", "Pretty sure."])
- def testFetchErrors(self) -> None:
- with requests.session() as s:
- with self.assertRaises(requests.HTTPError):
- paperdoorknob.fetch(
- f"http://localhost:{self._port}/not_found", s, TIMEOUT)
- with self.assertRaises(requests.HTTPError):
- paperdoorknob.fetch(
- f"http://localhost:{self._port}/server_error", s, TIMEOUT)
-
def testProcess(self) -> None:
- with requests.session() as s:
+ with DirectFetcher(TIMEOUT) as f:
buf = io.BytesIO()
paperdoorknob.process(
- f"http://localhost:{self._port}",
- s,
- TIMEOUT,
- buf,
- 'pandoc')
+ f"http://localhost:{self._port}", f, buf, 'pandoc')
self.assertEqual(buf.getvalue(), b'''\\documentclass{article}
\\begin{document}
This is glowfic
''')
def testPDF(self) -> None:
- with requests.session() as s:
+ with DirectFetcher(TIMEOUT) as f:
with open("test.tex", 'wb') as out:
paperdoorknob.process(
- f"http://localhost:{self._port}", s, TIMEOUT, out, 'pandoc')
+ f"http://localhost:{self._port}", f, out, 'pandoc')
subprocess.run(['pdflatex', 'test.tex'],
stdin=subprocess.DEVNULL, check=True)