]> git.scottworley.com Git - paperdoorknob/blobdiff - paperdoorknob.py
Strip links from meta fields
[paperdoorknob] / paperdoorknob.py
index 6a556753479b96f55c535f53a14d6c444822edf0..72ec1a5669f48595fba7b640aff98de2da1b6069 100644 (file)
@@ -35,12 +35,16 @@ def process(spec: Spec) -> None:
                           spec.geometry.encode('UTF-8') +
                           b']{geometry}\n')
     spec.texout.write(b'''\\begin{document}
-\\newcommand{\\href}[2]{#2\\footnote{#1}}
+\\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
         spec.log(f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)\r')
         spec.domfilter(r)