]> git.scottworley.com Git - paperdoorknob/commitdiff
Begin parsing glowfic html
authorScott Worley <scottworley@scottworley.com>
Fri, 24 Nov 2023 03:26:53 +0000 (19:26 -0800)
committerScott Worley <scottworley@scottworley.com>
Wed, 20 Dec 2023 01:37:10 +0000 (17:37 -0800)
paperdoorknob.py
paperdoorknob_test.py

index bb8bdd1ae38a660b3a6f90aa3d5aa317505beee3..b88c02d73e64a604763ad5c8ca8785055496d53f 100644 (file)
@@ -8,11 +8,24 @@
 from argparse import ArgumentParser
 import os.path
 from bs4 import BeautifulSoup
 from argparse import ArgumentParser
 import os.path
 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 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(
@@ -37,7 +50,8 @@ 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:
 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__':
index ab13eedf9730acccb9ec2b0f8ec64fc9c02ce016..83cf76ad4c9f17f7ac43150db9dbc3e62d7ea66e 100644 (file)
@@ -28,7 +28,13 @@ class FakeGlowficHTTPRequestHandler(BaseHTTPRequestHandler):
         return 200
 
     def do_GET(self) -> None:
         return 200
 
     def do_GET(self) -> None:
-        body = b'<html><body>This is glowfic</body></html>'
+        body = b'''<html>
+  <body>
+    <div class="post-container post-post">
+      This is glowfic
+    </div>
+  </body>
+</html>'''
         self.send_response(self._response_code())
         self.send_header("Content-type", "text/html")
         self.send_header("Content-Length", str(len(body)))
         self.send_response(self._response_code())
         self.send_header("Content-type", "text/html")
         self.send_header("Content-Length", str(len(body)))
@@ -83,11 +89,9 @@ class TestFetch(unittest.TestCase):
 
     def testFetchConents(self) -> None:
         with requests.session() as s:
 
     def testFetchConents(self) -> None:
         with requests.session() as s:
-            doc = paperdoorknob.fetch(
-                f"http://localhost:{self._port()}", s, TIMEOUT)
-            body = doc.body
-            assert body
-            self.assertEqual(body.text, "This is glowfic")
+            post = paperdoorknob.Post(paperdoorknob.fetch(
+                f"http://localhost:{self._port()}", s, TIMEOUT))
+            self.assertEqual(post.text().text.strip(), "This is glowfic")
 
     def testFetchErrors(self) -> None:
         with requests.session() as s:
 
     def testFetchErrors(self) -> None:
         with requests.session() as s: