Texifier interface

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index f83c9f869651a0a35fd86fdf11fdb004e5e5ae4d..3b86cb8da08144c3862ab80b757b0194c086dbad 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -8,7 +8,6 @@
 from argparse import ArgumentParser
 import itertools
 import os.path
-import subprocess
 
 from typing import IO, Iterable
 
@@ -17,6 +16,7 @@ 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:
@@ -40,8 +40,8 @@ def command_line_parser() -> ArgumentParser:
     return parser
 
 
-def fetch(url: str, fetcher: Fetcher) -> BeautifulSoup:
-    return BeautifulSoup(fetcher.fetch(url), 'html.parser')
+def parse(content: bytes) -> BeautifulSoup:
+    return BeautifulSoup(content, 'html.parser')
 
 
 def clean(html: BeautifulSoup) -> BeautifulSoup:
@@ -68,30 +68,24 @@ def replies(html: BeautifulSoup) -> Iterable[Tag]:
     return itertools.chain([text()], the_replies())
 
 
-def html_to_tex(pandoc: str, tag: Tag) -> bytes:
-    return subprocess.run([pandoc, '--from=html', '--to=latex'],
-                          input=tag.encode(),
-                          stdout=subprocess.PIPE,
-                          check=True).stdout
-
-
 def process(
         url: str,
         fetcher: Fetcher,
-        texout: IO[bytes],
-        pandoc: str) -> None:
+        texifier: Texifier,
+        texout: IO[bytes]) -> None:
     texout.write(b'\\documentclass{article}\n\\begin{document}\n')
-    html = clean(fetch(url, fetcher))
+    html = clean(parse(fetcher.fetch(url)))
     for r in replies(html):
-        texout.write(html_to_tex(pandoc, r))
+        texout.write(texifier.texify(r))
     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, texout, args.pandoc or 'pandoc')
+            process(args.url, fetcher, texifier, texout)
 
 
 if __name__ == '__main__':