X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/136277e30143cd1219c896bb4980027ac3c6dbe1..170e50a0efa18ed0f5f2ff644793469894983029:/paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index bb8bdd1..fb2df1f 100644
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -6,13 +6,42 @@
 
 
 from argparse import ArgumentParser
+import itertools
 import os.path
+
+from typing import Iterable
+
 from bs4 import BeautifulSoup
+from bs4.element import Tag
 import requests
 import requests_cache
 from xdg_base_dirs import xdg_cache_home
 
 
+class Post:
+    def __init__(self, html: BeautifulSoup) -> None:
+        self._html = html
+        for eb in self._html.find_all("div", class_="post-edit-box"):
+            eb.decompose()
+        for footer in self._html.find_all("div", class_="post-footer"):
+            footer.decompose()
+
+    def text(self) -> Tag:
+        body = self._html.body
+        assert body
+        text = body.find_next("div", class_="post-post")
+        assert isinstance(text, Tag)
+        return text
+
+    def replies(self) -> Iterable[Tag]:
+        replies = self._html.find_all("div", class_="post-reply")
+        assert all(isinstance(r, Tag) for r in replies)
+        return replies
+
+    def entries(self) -> Iterable[Tag]:
+        return itertools.chain([self.text()], self.replies())
+
+
 def command_line_parser() -> ArgumentParser:
     parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
     parser.add_argument(
@@ -37,7 +66,8 @@ def fetch(url: str, session: requests.Session, timeout: int) -> BeautifulSoup:
 def main() -> None:
     args = command_line_parser().parse_args()
     with requests_cache.CachedSession(args.cache_path, cache_control=True) as session:
-        fetch(args.url, session, args.timeout)
+        html = fetch(args.url, session, args.timeout)
+        Post(html)
 
 
 if __name__ == '__main__':