Render links as footnotes

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index f8b336194974668477b5739b74c014a00d7e7549..6a556753479b96f55c535f53a14d6c444822edf0 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
 
 
@@ -16,21 +17,36 @@ 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)))
-    for r in chunkDOMs(html):
+    spec.texout.write(b'''\\begin{document}
+\\newcommand{\\href}[2]{#2\\footnote{#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)):
+        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')