Always have Thread.__init__ fetch the HTML

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index 81831360e462a6a9af1b366f3384e566e0adb412..d545f6184514f0386068871a247b6a30fe1aad18 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -6,17 +6,11 @@
 
 from typing import Any, Iterable
 
-from bs4 import BeautifulSoup
-
 from args import spec_from_commandline_args
-from glowfic import chunkDOMs, flatURL, makeChunk
+from glowfic import makeChunk, renderChunk, Thread
 from spec import Spec
 
 
-def parse(content: bytes) -> BeautifulSoup:
-    return BeautifulSoup(content, 'html.parser')
-
-
 def ilen(it: Iterable[Any]) -> int:
     return sum(1 for _ in it)
 
@@ -37,20 +31,23 @@ def process(spec: Spec) -> None:
     spec.texout.write(br'''\begin{document}
 \newcommand{\href}[2]{#2\footnote{\detokenize{#1}}}
 \def\glowiconsize{%fmm}
-''' % spec.icon_size)
-    url = flatURL(spec.url)
-    spec.log('Fetching HTML...\r')
-    html = spec.fetcher.fetch(url)
-    spec.log('Parsing HTML...\r')
-    dom = parse(spec.htmlfilter(html))
+\newcommand{\glowicon}[1]{\includegraphics[
+  width=\glowiconsize,height=\glowiconsize,keepaspectratio
+]{#1}}
+\newcommand{\ifnotempty}[2]{\expandafter\ifx\expandafter\relax
+  \detokenize{#1}\relax\else #2\fi}
+%s
+''' % (spec.icon_size, spec.layout))
+    thread = Thread(spec)
     spec.log('Counting chunks...\r')
-    num_chunks = ilen(chunkDOMs(dom))
-    for i, r in enumerate(chunkDOMs(dom)):
+    num_chunks = ilen(thread.chunkDOMs())
+    title = thread.title() or "chunk"
+    for i, r in enumerate(thread.chunkDOMs()):
         percent = 100.0 * i / num_chunks
-        spec.log(f'Processing chunk {i} of {num_chunks} ({percent:.1f}%)\r')
+        spec.log(f'Processing {title} {i} of {num_chunks} ({percent:.1f}%)\r')
         spec.domfilter(r)
         chunk = makeChunk(r, spec.images)
-        spec.texout.write(spec.texfilter(spec.layout.renderChunk(chunk)))
+        spec.texout.write(spec.texfilter(renderChunk(spec.texifier, chunk)))
     spec.log('')
     spec.texout.write(b'\\end{document}\n')