]> git.scottworley.com Git - paperdoorknob/blobdiff - paperdoorknob.py
Strip edit-boxes and footers
[paperdoorknob] / paperdoorknob.py
index b88c02d73e64a604763ad5c8ca8785055496d53f..fb2df1f1cc1c7443bc6e07018bad7148eb12a5c6 100644 (file)
@@ -6,7 +6,11 @@
 
 
 from argparse import ArgumentParser
+import itertools
 import os.path
+
+from typing import Iterable
+
 from bs4 import BeautifulSoup
 from bs4.element import Tag
 import requests
@@ -17,6 +21,10 @@ 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
@@ -25,6 +33,14 @@ class 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')