Move per-thread processing stuff into Thread

[paperdoorknob] / args.py

diff --git a/args.py b/args.py
index 2a03836854777b6ce3a6b7b1d4171923508b7e0e..de6fc01f58d353c26cb245352640a0ad1970cc0b 100644 (file)
--- a/args.py
+++ b/args.py
@@ -8,40 +8,105 @@
 from argparse import ArgumentParser
 from contextlib import contextmanager
 import os.path
 from argparse import ArgumentParser
 from contextlib import contextmanager
 import os.path

+from sys import stderr

 
 from typing import Iterator
 
 from xdg_base_dirs import xdg_cache_home
 
 
 from typing import Iterator
 
 from xdg_base_dirs import xdg_cache_home
 

+from domfilter import ApplyDOMFilters, DOMFilters

 from fetch import CachingFetcher
 from fetch import CachingFetcher

+from glowfic import BesideIconLayout, BelowIconLayout
+from htmlfilter import ApplyHTMLFilters, HTMLFilters
+from texfilter import ApplyTexFilters, TexFilters
+from images import DiskImageStore

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

+def _print_status(msg: str) -> None:
+    print(msg, file=stderr, end='' if msg.endswith('\r') else '\n')
+
+

 def _command_line_parser() -> ArgumentParser:
     parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
     parser.add_argument(
 def _command_line_parser() -> ArgumentParser:
     parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
     parser.add_argument(

-        '--cache_path',
+        '--cache-path',

         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(
+        '--geometry',
+        help='''Page size and margin control
+See https://faculty.bard.edu/bloch/geometry.pdf for details
+(default: %(default)s)''',
+        default='paperwidth=5.5in,paperheight=8.5in,nohead,' +
+        'tmargin=15mm,hmargin=15mm,bmargin=17mm,foot=4mm')
+    parser.add_argument(
+        '--htmlfilters',
+        help='Which HTML filters to use (default: %(default)s)',
+        default=','.join(f[0] for f in HTMLFilters))
+    parser.add_argument(
+        '--image-size',
+        help='How large the icon images are, in mm',
+        default=20)
+    parser.add_argument(
+        '--layout',
+        default='beside',
+        help='Whether to put character and author information `beside` or `below` the icon ' +
+             '(default: beside)')

     parser.add_argument(
         '--out',
         help='The filename stem at which to write output ' +
              '(eg: "%(default)s" produces %(default)s.tex, %(default)s.pdf, etc.)',
         default='book')
     parser.add_argument('--pandoc', help='Location of the pandoc executable')
     parser.add_argument(
         '--out',
         help='The filename stem at which to write output ' +
              '(eg: "%(default)s" produces %(default)s.tex, %(default)s.pdf, etc.)',
         default='book')
     parser.add_argument('--pandoc', help='Location of the pandoc executable')

+    parser.add_argument(
+        '--quiet',
+        action='store_true',
+        help='Suppress progress messages')
+    parser.add_argument(
+        '--texfilters',
+        help='Which TeX filters to use (default: %(default)s)',
+        default=','.join(f[0] for f in TexFilters))

     parser.add_argument(
         '--timeout',
         help='How long to wait for HTTP requests, in seconds',
         default=30)
     parser.add_argument(
         '--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
 
 
 @contextmanager
 def spec_from_commandline_args() -> Iterator[Spec]:
     args = _command_line_parser().parse_args()
     return parser
 
 
 @contextmanager
 def spec_from_commandline_args() -> Iterator[Spec]:
     args = _command_line_parser().parse_args()

-    with CachingFetcher(args.cache_path, args.timeout) as fetcher:
+    texifier = PandocTexifier(args.pandoc or 'pandoc')
+    layout: bytes
+    if args.layout == 'below':
+        layout = BelowIconLayout
+    elif args.layout == 'beside':
+        layout = BesideIconLayout
+    else:
+        raise ValueError(f'Unknown layout: {args.layout}')
+    log = (lambda _: None) if args.quiet else _print_status
+    with CachingFetcher(args.cache_path, args.timeout, log) as fetcher:

         with open(args.out + '.tex', 'wb') as texout:
         with open(args.out + '.tex', 'wb') as texout:

-            yield Spec(args.url, fetcher, PandocTexifier(args.pandoc or 'pandoc'), texout)
+            yield Spec(
+                args.url,
+                fetcher,
+                DiskImageStore(args.out + '_images', fetcher),
+                lambda x: ApplyHTMLFilters(args.htmlfilters, x),
+                lambda x: ApplyDOMFilters(args.domfilters, x),
+                texifier,
+                lambda x: ApplyTexFilters(args.texfilters, x),
+                args.image_size,
+                layout,
+                args.geometry,
+                texout,
+                log)