Rename: fetch → parse

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index 32daf4f4d9372b07ee1e5eb801037ea2efa129ad..31ce02c494ea29aadc8794d75f11ee8397729a2c 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -14,10 +14,10 @@ from typing import IO, Iterable
 
 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')
@@ -40,10 +40,8 @@ def command_line_parser() -> ArgumentParser:
     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 parse(content: bytes) -> BeautifulSoup:
+    return BeautifulSoup(content, 'html.parser')
 
 
 def clean(html: BeautifulSoup) -> BeautifulSoup:
@@ -79,26 +77,21 @@ def html_to_tex(pandoc: str, tag: Tag) -> bytes:
 
 def process(
         url: str,
-        session: requests.Session,
-        timeout: int,
+        fetcher: Fetcher,
         texout: IO[bytes],
-        pandoc: str,
-) -> None:
-    html = clean(fetch(url, session, timeout))
+        pandoc: str) -> None:
+    texout.write(b'\\documentclass{article}\n\\begin{document}\n')
+    html = clean(parse(fetcher.fetch(url)))
     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__':