Project Lawful start URL in --help

[paperdoorknob] / args.py

diff --git a/args.py b/args.py
index 1931d4271f242dc918287d28a7bedf2e4ee71f81..fb36083b28b9bcf852d37963f8b53c33760819d3 100644 (file)
--- a/args.py
+++ b/args.py
@@ -13,6 +13,7 @@ from typing import Iterator
 
 from xdg_base_dirs import xdg_cache_home
 
 
 from xdg_base_dirs import xdg_cache_home
 

+from domfilter import ApplyDOMFilters, DOMFilters

 from fetch import CachingFetcher
 from htmlfilter import ApplyHTMLFilters, HTMLFilters
 from spec import Spec
 from fetch import CachingFetcher
 from htmlfilter import ApplyHTMLFilters, HTMLFilters
 from spec import Spec


@@ -26,6 +27,10 @@ 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(
+        '--domfilters',
+        help='Which DOM filters to use (default: %(default)s)',
+        default=','.join(f[0] for f in DOMFilters))

     parser.add_argument(
         '--htmlfilters',
         help='Which HTML filters to use (default: %(default)s)',
     parser.add_argument(
         '--htmlfilters',
         help='Which HTML filters to use (default: %(default)s)',


@@ -40,7 +45,9 @@ def _command_line_parser() -> ArgumentParser:
         '--timeout',
         help='How long to wait for HTTP requests, in seconds',
         default=30)
         '--timeout',
         help='How long to wait for HTTP requests, in seconds',
         default=30)

-    parser.add_argument('url', help='URL to retrieve')
+    parser.add_argument(
+        'url',
+        help='URL to retrieve (example: https://www.projectlawful.com/posts/4582 )')

     return parser
 
 
     return parser
 
 


@@ -53,5 +60,6 @@ def spec_from_commandline_args() -> Iterator[Spec]:
                 args.url,
                 fetcher,
                 lambda x: ApplyHTMLFilters(args.htmlfilters, x),
                 args.url,
                 fetcher,
                 lambda x: ApplyHTMLFilters(args.htmlfilters, x),

+                lambda x: ApplyDOMFilters(args.domfilters, x),

                 PandocTexifier(args.pandoc or 'pandoc'),
                 texout)
                 PandocTexifier(args.pandoc or 'pandoc'),
                 texout)