Support _ in URLs

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index 26a60fbc0524f5daf46fbc8f92e131c011e0fef1..72ec1a5669f48595fba7b640aff98de2da1b6069 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -4,8 +4,6 @@
 # under the terms of the GNU General Public License as published by the
 # Free Software Foundation, version 3.
 
-from sys import stderr
-
 from typing import Any, Iterable
 
 from bs4 import BeautifulSoup
@@ -36,19 +34,23 @@ def process(spec: Spec) -> None:
         spec.texout.write(b'\\usepackage[' +
                           spec.geometry.encode('UTF-8') +
                           b']{geometry}\n')
-    spec.texout.write(b'\\begin{document}\n')
+    spec.texout.write(b'''\\begin{document}
+\\newcommand{\\href}[2]{#2\\footnote{\\detokenize{#1}}}
+''')
     url = flatURL(spec.url)
-    html = parse(spec.htmlfilter(spec.fetcher.fetch(url)))
-    num_chunks = ilen(chunkDOMs(html))
-    for i, r in enumerate(chunkDOMs(html)):
+    spec.log('Fetching HTML...\r')
+    html = spec.fetcher.fetch(url)
+    spec.log('Parsing HTML...\r')
+    dom = parse(spec.htmlfilter(html))
+    spec.log('Counting chunks...\r')
+    num_chunks = ilen(chunkDOMs(dom))
+    for i, r in enumerate(chunkDOMs(dom)):
         percent = 100.0 * i / num_chunks
-        print(
-            f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)',
-            end='\r',
-            file=stderr)
+        spec.log(f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)\r')
         spec.domfilter(r)
         chunk = makeChunk(r, spec.images)
         spec.texout.write(spec.texfilter(spec.layout.renderChunk(chunk)))
+    spec.log('')
     spec.texout.write(b'\\end{document}\n')