X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/b254976424d0f56fab14f8897dade345ce46a2d1..551bb1c9f85382a90abc89ac622bd363dccd36a7:/paperdoorknob.py?ds=sidebyside diff --git a/paperdoorknob.py b/paperdoorknob.py index 6a55675..0fd1fde 100644 --- 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)