Only look within each chunk-dom for chunk fields

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index 6a556753479b96f55c535f53a14d6c444822edf0..0fd1fdeceb6a5b9ff64b1a421fcc3b8f71a57c5a 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -38,9 +38,13 @@ def process(spec: Spec) -> None:
 \\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)