Support _ in URLs

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index 24ccccd19d471482cc37542731c1bc1699bd6b41..72ec1a5669f48595fba7b640aff98de2da1b6069 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -4,11 +4,12 @@
 # under the terms of the GNU General Public License as published by the
 # Free Software Foundation, version 3.
 
+from typing import Any, Iterable
 
 from bs4 import BeautifulSoup
 
 from args import spec_from_commandline_args
-from glowfic import chunkDOMs, makeChunk
+from glowfic import chunkDOMs, flatURL, makeChunk
 from spec import Spec
 
 
@@ -16,30 +17,40 @@ def parse(content: bytes) -> BeautifulSoup:
     return BeautifulSoup(content, 'html.parser')
 
 
+def ilen(it: Iterable[Any]) -> int:
+    return sum(1 for _ in it)
+
+
 def process(spec: Spec) -> None:
     spec.texout.write(b'''\\documentclass{article}
+\\usepackage{booktabs}
 \\usepackage{graphicx}
-\\usepackage{wrapfig}
+\\usepackage{longtable}
+\\usepackage{soul}
+\\usepackage{varwidth}
+\\usepackage{wrapstuff}
 ''')
     if spec.geometry is not None:
         spec.texout.write(b'\\usepackage[' +
                           spec.geometry.encode('UTF-8') +
                           b']{geometry}\n')
-    spec.texout.write(b'\\begin{document}\n')
-    html = parse(spec.htmlfilter(spec.fetcher.fetch(spec.url)))
-    first = True
-    for r in chunkDOMs(html):
-        if first:
-            first = False
-        else:
-            # h/t https://tex.stackexchange.com/questions/309856
-            spec.texout.write(b'''
-\\vspace{-.5\\ht\\strutbox}\\noindent\\hrulefill
-
+    spec.texout.write(b'''\\begin{document}
+\\newcommand{\\href}[2]{#2\\footnote{\\detokenize{#1}}}
 ''')
+    url = flatURL(spec.url)
+    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
+        spec.log(f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)\r')
         spec.domfilter(r)
         chunk = makeChunk(r, spec.images)
-        spec.texout.write(spec.layout.renderChunk(chunk))
+        spec.texout.write(spec.texfilter(spec.layout.renderChunk(chunk)))
+    spec.log('')
     spec.texout.write(b'\\end{document}\n')