]> git.scottworley.com Git - paperdoorknob/commitdiff
Progress indicator
authorScott Worley <scottworley@scottworley.com>
Sat, 30 Dec 2023 00:06:38 +0000 (16:06 -0800)
committerScott Worley <scottworley@scottworley.com>
Sat, 30 Dec 2023 00:06:38 +0000 (16:06 -0800)
paperdoorknob.py

index 0da6d97c8626938e29c407dd507467897994c9ea..26a60fbc0524f5daf46fbc8f92e131c011e0fef1 100644 (file)
@@ -4,6 +4,9 @@
 # under the terms of the GNU General Public License as published by the
 # Free Software Foundation, version 3.
 
 # under the terms of the GNU General Public License as published by the
 # Free Software Foundation, version 3.
 
+from sys import stderr
+
+from typing import Any, Iterable
 
 from bs4 import BeautifulSoup
 
 
 from bs4 import BeautifulSoup
 
@@ -16,6 +19,10 @@ def parse(content: bytes) -> BeautifulSoup:
     return BeautifulSoup(content, 'html.parser')
 
 
     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}
 def process(spec: Spec) -> None:
     spec.texout.write(b'''\\documentclass{article}
 \\usepackage{booktabs}
@@ -32,7 +39,13 @@ def process(spec: Spec) -> None:
     spec.texout.write(b'\\begin{document}\n')
     url = flatURL(spec.url)
     html = parse(spec.htmlfilter(spec.fetcher.fetch(url)))
     spec.texout.write(b'\\begin{document}\n')
     url = flatURL(spec.url)
     html = parse(spec.htmlfilter(spec.fetcher.fetch(url)))
-    for r in chunkDOMs(html):
+    num_chunks = ilen(chunkDOMs(html))
+    for i, r in enumerate(chunkDOMs(html)):
+        percent = 100.0 * i / num_chunks
+        print(
+            f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)',
+            end='\r',
+            file=stderr)
         spec.domfilter(r)
         chunk = makeChunk(r, spec.images)
         spec.texout.write(spec.texfilter(spec.layout.renderChunk(chunk)))
         spec.domfilter(r)
         chunk = makeChunk(r, spec.images)
         spec.texout.write(spec.texfilter(spec.layout.renderChunk(chunk)))