Only look within each chunk-dom for chunk fields

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index 4e7cdf6840b8bbd7fb6a227db538bc9db9cde915..0fd1fdeceb6a5b9ff64b1a421fcc3b8f71a57c5a 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -34,11 +34,17 @@ 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{#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
         spec.log(f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)\r')
         spec.domfilter(r)