X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/92b11a100bc251b0b13d193b77f25b5f0bce08e7..386218397a4723129b31e3954bb960da61f72474:/paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index 8f1d840..653d0c6 100644
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -5,16 +5,56 @@
 # Free Software Foundation, version 3.
 
 
-from argparse import ArgumentParser
+import itertools
 
+from typing import Iterable
 
-def command_line_parser() -> ArgumentParser:
-    parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
-    return parser
+from bs4 import BeautifulSoup
+from bs4.element import Tag
+
+from args import spec_from_commandline_args
+from spec import Spec
+
+
+def parse(content: bytes) -> BeautifulSoup:
+    return BeautifulSoup(content, 'html.parser')
+
+
+def clean(html: BeautifulSoup) -> BeautifulSoup:
+    for eb in html.find_all("div", class_="post-edit-box"):
+        eb.decompose()
+    for footer in html.find_all("div", class_="post-footer"):
+        footer.decompose()
+    return html
+
+
+def replies(html: BeautifulSoup) -> Iterable[Tag]:
+    def text() -> Tag:
+        body = html.body
+        assert body
+        text = body.find_next("div", class_="post-post")
+        assert isinstance(text, Tag)
+        return text
+
+    def the_replies() -> Iterable[Tag]:
+        rs = html.find_all("div", class_="post-reply")
+        assert all(isinstance(r, Tag) for r in rs)
+        return rs
+
+    return itertools.chain([text()], the_replies())
+
+
+def process(spec: Spec) -> None:
+    spec.texout.write(b'\\documentclass{article}\n\\begin{document}\n')
+    html = clean(parse(spec.fetcher.fetch(spec.url)))
+    for r in replies(html):
+        spec.texout.write(spec.texifier.texify(r))
+    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__':