Fix cleaning text with character references

You need to call `.close()` to force the parser to flush all its
buffered output.

cry/feed.py

@@ -284,6 +284,7 @@ def clean_text(text: str) -> str:
     writer = io.StringIO()
     cleaner = Cleaner(writer)
     cleaner.feed(text)
+    cleaner.close()
     return MULTI_SPACES.sub(" ", writer.getvalue())
 
 

tests/test_feed.py

@@ -1,6 +1,7 @@
 import asyncio
 import dataclasses
 import http.server
+import pathlib
 import threading
 import typing
 
@@ -89,6 +90,22 @@ class TestWebServer:
 
         self.respond(path, MockResponse(code, headers, content))
 
+    def handle_canned_response(self, path: str, response_path: str):
+        headers = []
+        with open(pathlib.Path(__file__).parent / response_path, "rb") as f:
+            _, code = f.readline().decode("utf-8").split(" ", maxsplit=1)
+            while True:
+                line = f.readline().decode("utf-8").rstrip()
+                if len(line) == 0:
+                    break
+                header = tuple(line.split(":", maxsplit=1))
+                assert len(header) == 2
+                headers.append(header)
+
+            content = f.read()
+
+        self.respond(path, MockResponse(int(code), headers, content))
+
     def respond(self, path: str, response: MockResponse):
         response_list = self.handlers.get(path)
         if response_list is None:
@@ -198,3 +215,14 @@ def test_fetch_after_permanent_to_permanent_redirect():
         # NOTE: we should record the latest redirect.
         assert new_meta.url == server.make_url("/two")
         assert isinstance(result, feed.Feed)
+
+
+def test_titles_with_entities():
+    with TestWebServer() as server:
+        server.handle_canned_response("/feed", "feeds/steveklabnik.com.xml")
+
+        meta = feed.FeedMeta.from_url(server.make_url("/feed"))
+        result, _ = asyncio.run(feed.fetch_feed(meta))
+
+        assert isinstance(result, feed.Feed)
+        assert result.entries[0].title == "When should I use String vs &str?"