]> git.scottworley.com Git - paperdoorknob/blobdiff - paperdoorknob.py
wrapfig → wrapstuff, for \wrapstuffclear
[paperdoorknob] / paperdoorknob.py
index b4f5e52417ba1f935bfb8b128f0b7df2da6483e7..f2a81d2247b898f7450f2e9d037e976db3821762 100644 (file)
@@ -5,29 +5,47 @@
 # 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, session: requests.Session, timeout: int) -> None:
-    with session.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, 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{graphicx}
+\\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')
+    html = parse(spec.htmlfilter(spec.fetcher.fetch(spec.url)))
+    first = True
+    for r in chunkDOMs(html):
+        if first:
+            first = False
+        else:
+            # h/t https://tex.stackexchange.com/questions/309856
+            spec.texout.write(b'''
+\\wrapstuffclear\\vspace{-.5\\ht\\strutbox}\\noindent\\hrulefill
+
+''')
+        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:
-    args = command_line_parser().parse_args()
-    with requests.session() as session:
-        fetch(args.url, session, args.timeout)
+    with spec_from_commandline_args() as spec:
+        process(spec)
 
 
 if __name__ == '__main__':