Replies

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index b7e4349b98ab1d97ae2ff0ad6f50e4740bb6440f..0c6ec856b50377f1b895a2dfe2d9681badf93dba 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -7,11 +7,33 @@
 
 from argparse import ArgumentParser
 import os.path
 
 from argparse import ArgumentParser
 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
 
 
 import requests
 import requests_cache
 from xdg_base_dirs import xdg_cache_home
 
 

+class Post:
+    def __init__(self, html: BeautifulSoup) -> None:
+        self._html = html
+
+    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 command_line_parser() -> ArgumentParser:
     parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
     parser.add_argument(
 def command_line_parser() -> ArgumentParser:
     parser = ArgumentParser(prog='paperdoorknob', description='Print glowfic')
     parser.add_argument(


@@ -27,15 +49,17 @@ def command_line_parser() -> ArgumentParser:
     return parser
 
 
     return parser
 
 

-def fetch(url: str, session: requests.Session, timeout: int) -> None:
+def fetch(url: str, session: requests.Session, timeout: int) -> BeautifulSoup:

     with session.get(url, timeout=timeout) as r:
         r.raise_for_status()
     with session.get(url, timeout=timeout) as r:
         r.raise_for_status()

+        return BeautifulSoup(r.text, 'html.parser')

 
 
 def main() -> None:
     args = command_line_parser().parse_args()
     with requests_cache.CachedSession(args.cache_path, cache_control=True) as session:
 
 
 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__':
 
 
 if __name__ == '__main__':