Strip all  

[paperdoorknob] / args.py

diff --git a/args.py b/args.py
index 2a03836854777b6ce3a6b7b1d4171923508b7e0e..1931d4271f242dc918287d28a7bedf2e4ee71f81 100644 (file)
--- a/args.py
+++ b/args.py
@@ -14,6 +14,7 @@ from typing import Iterator
 from xdg_base_dirs import xdg_cache_home
 
 from fetch import CachingFetcher
 from xdg_base_dirs import xdg_cache_home
 
 from fetch import CachingFetcher

+from htmlfilter import ApplyHTMLFilters, HTMLFilters

 from spec import Spec
 from texify import PandocTexifier
 
 from spec import Spec
 from texify import PandocTexifier
 


@@ -25,6 +26,10 @@ def _command_line_parser() -> ArgumentParser:
         metavar='PATH',
         help='Where to keep the http cache (instead of %(default)s)',
         default=os.path.join(xdg_cache_home(), "paperdoorknob"))
         metavar='PATH',
         help='Where to keep the http cache (instead of %(default)s)',
         default=os.path.join(xdg_cache_home(), "paperdoorknob"))

+    parser.add_argument(
+        '--htmlfilters',
+        help='Which HTML filters to use (default: %(default)s)',
+        default=','.join(f[0] for f in HTMLFilters))

     parser.add_argument(
         '--out',
         help='The filename stem at which to write output ' +
     parser.add_argument(
         '--out',
         help='The filename stem at which to write output ' +


@@ -44,4 +49,9 @@ 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:
     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)
+            yield Spec(
+                args.url,
+                fetcher,
+                lambda x: ApplyHTMLFilters(args.htmlfilters, x),
+                PandocTexifier(args.pandoc or 'pandoc'),
+                texout)