Use LaTeX packages: longtable and booktabs

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index c684739e0a363ee310e27343a08effde2bb4dc89..b5b7a44a433f2410d64fb12c021348fac25171b8 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -5,29 +5,42 @@
 # Free Software Foundation, version 3.
 
 
 # Free Software Foundation, version 3.
 
 

-from argparse import ArgumentParser
-import requests
-
-
-def command_line_parser() -> ArgumentParser:
-    parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
-    parser.add_argument(
-        '--timeout',
-        help='How long to wait for HTTP requests, in seconds',
-        default=30)
-    parser.add_argument('url', help='URL to retrieve')
-    return parser
-
-
-def fetch(url: str, timeout: int) -> None:
-    with requests.session() as s:
-        with s.get(url, timeout=timeout) as r:
-            r.raise_for_status()
+from bs4 import BeautifulSoup
+
+from args import spec_from_commandline_args
+from glowfic import chunkDOMs, flatURL, makeChunk
+from spec import Spec
+
+
+def parse(content: bytes) -> BeautifulSoup:
+    return BeautifulSoup(content, 'html.parser')
+
+
+def process(spec: Spec) -> None:
+    spec.texout.write(b'''\\documentclass{article}
+\\usepackage{booktabs}
+\\usepackage{graphicx}
+\\usepackage{longtable}
+\\usepackage{varwidth}
+\\usepackage{wrapstuff}
+''')
+    if spec.geometry is not None:
+        spec.texout.write(b'\\usepackage[' +
+                          spec.geometry.encode('UTF-8') +
+                          b']{geometry}\n')
+    spec.texout.write(b'\\begin{document}\n')
+    url = flatURL(spec.url)
+    html = parse(spec.htmlfilter(spec.fetcher.fetch(url)))
+    for r in chunkDOMs(html):
+        spec.domfilter(r)
+        chunk = makeChunk(r, spec.images)
+        spec.texout.write(spec.layout.renderChunk(chunk))
+    spec.texout.write(b'\\end{document}\n')

 
 
 def main() -> None:
 
 
 def main() -> None:

-    args = command_line_parser().parse_args()
-    fetch(args.url, args.timeout)
+    with spec_from_commandline_args() as spec:
+        process(spec)

 
 
 if __name__ == '__main__':
 
 
 if __name__ == '__main__':