]> git.scottworley.com Git - paperdoorknob/blobdiff - paperdoorknob.py
texfilter to work around \emph nesting issue
[paperdoorknob] / paperdoorknob.py
index 8f1d840d4a4f0bc9155e6e645b19535442cb2914..b364f4d63653d3fa486ca764bb60768967b60763 100644 (file)
@@ -5,16 +5,42 @@
 # Free Software Foundation, version 3.
 
 
-from argparse import ArgumentParser
-
-
-def command_line_parser() -> ArgumentParser:
-    parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
-    return parser
+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.texfilter(spec.layout.renderChunk(chunk)))
+    spec.texout.write(b'\\end{document}\n')
 
 
 def main() -> None:
-    command_line_parser().parse_args()
+    with spec_from_commandline_args() as spec:
+        process(spec)
 
 
 if __name__ == '__main__':