feat: Ignore quotes and double quotes in filtered words and categories

kevinrenskers · kevinrenskers · commit 202b9ff8436b · 2025-04-28T15:27:02.000+02:00
diff --git a/example/uv.lock b/example/uv.lock
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "django-rss-filter"
-version = "0.10.1"
+version = "0.11.0"
 description = "Filter public RSS feeds, remove articles that contain certain keywords or categories."
 authors = [
     {name = "Kevin Renskers", email = "kevin@loopwerk.io"},
diff --git a/rssfilter/utils.py b/rssfilter/utils.py
@@ -29,20 +29,26 @@ def validate_feed(feed_url: str) -> FeedValidationSuccess | FeedValidationError:
         # that validating and then fetching the feed is done in a consistent
         # manner.
         r = httpx.get(feed_url, follow_redirects=True, timeout=2, headers={"User-Agent": USER_AGENT})
-
-        feed = feedparser.parse(r.text)
-        version = feed.get("version", "")
-        if not version:
+        feed = validate_feed_body(r.text)
+        if not feed:
             return FeedValidationError(False, "This doesn't seem to be a valid RSS or Atom feed")
         return FeedValidationSuccess(True, feed)
-    except ValueError:
-        return FeedValidationError(False, "This doesn't seem to be a valid RSS or Atom feed")
     except ConnectTimeout:
         return FeedValidationError(False, "Couldn't load the URL due to a connection timeout")
     except ConnectError:
         return FeedValidationError(False, "Couldn't load the URL due to a connection error")
 
 
+def validate_feed_body(body: str) -> Literal[False] | FeedParserDict:
+    try:
+        feed = feedparser.parse(body)
+        return feed if feed.version else False
+    except AttributeError:
+        return False
+    except ValueError:
+        return False
+
+
 def filter_feed(feed_body: str, filtered_words: str, filtered_categories: str) -> str:
     feed = feedparser.parse(feed_body)
 
@@ -59,8 +65,12 @@ def filter_feed(feed_body: str, filtered_words: str, filtered_categories: str) -
     if published := feed.feed.get("published"):
         fg.pubDate(published)
 
-    filtered_words_list = [item.strip().lower() for item in filtered_words.split(",") if item.strip()]
-    filtered_categories_list = [item.strip().lower() for item in filtered_categories.split(",") if item.strip()]
+    filtered_words_list = [item.lower().strip().strip('"').strip("'") for item in filtered_words.split(",")]
+    filtered_categories_list = [item.lower().strip().strip('"').strip("'") for item in filtered_categories.split(",")]
+
+    # Filter out empty strings
+    filtered_words_list = [x for x in filtered_words_list if x]
+    filtered_categories_list = [x for x in filtered_categories_list if x]
 
     for entry in feed.entries:
         # Check if the title contains filtered words
diff --git a/tests/test_models.py b/tests/test_models.py
@@ -0,0 +1,21 @@
+from django.test import TestCase
+from django.utils import timezone
+
+from rssfilter.models import FilteredFeed
+
+
+class ModelsTest(TestCase):
+    def test_clear_cache_on_safe(self):
+        feed = FilteredFeed.objects.create(
+            feed_url="http://www.example.com/",
+            cache_date=timezone.now(),
+            filtered_feed_body="CACHED",
+        )
+
+        self.assertEqual(feed.filtered_feed_body, "CACHED")
+
+        feed.filtered_words = "Changed value"
+        feed.save()
+
+        self.assertEqual(feed.cache_date, None)
+        self.assertEqual(feed.filtered_feed_body, "")
diff --git a/tests/test_utils.py b/tests/test_utils.py
@@ -1,7 +1,7 @@
 import feedparser
 from django.test import TestCase
 
-from rssfilter.utils import filter_feed
+from rssfilter.utils import filter_feed, validate_feed_body
 
 body = """
 <?xml version="1.0" encoding="UTF-8"?>
@@ -35,6 +35,18 @@
 
 
 class UtilsTest(TestCase):
+    def test_validate_feed_body_vlaid(self):
+        result = validate_feed_body(body)
+        self.assertTrue(result)
+
+    def test_validate_feed_body_invalid(self):
+        result = validate_feed_body("<html>Hello</html>")
+        self.assertFalse(result)
+
+    def test_validate_feed_body_empty(self):
+        result = validate_feed_body("")
+        self.assertFalse(result)
+
     def test_filter_words(self):
         filtered_feed_body = filter_feed(body, filtered_words="One", filtered_categories="")
         filtered_feed = feedparser.parse(filtered_feed_body)
@@ -55,6 +67,12 @@ def test_filter_words_empty(self):
 
         self.assertEqual(len(filtered_feed.entries), 2)
 
+    def test_filter_words_with_empty_quotes(self):
+        filtered_feed_body = filter_feed(body, filtered_words='""', filtered_categories="")
+        filtered_feed = feedparser.parse(filtered_feed_body)
+
+        self.assertEqual(len(filtered_feed.entries), 2)
+
     def test_filter_words_not_found(self):
         filtered_feed_body = filter_feed(body, filtered_words="Foo", filtered_categories="")
         filtered_feed = feedparser.parse(filtered_feed_body)
@@ -68,6 +86,20 @@ def test_filter_words_comma_seperated(self):
         self.assertEqual(len(filtered_feed.entries), 1)
         self.assertEqual(filtered_feed.entries[0].title, "Article Two")
 
+    def test_filter_words_with_quotes(self):
+        filtered_feed_body = filter_feed(body, filtered_words="'Foo', 'Bar', 'One'", filtered_categories="")
+        filtered_feed = feedparser.parse(filtered_feed_body)
+
+        self.assertEqual(len(filtered_feed.entries), 1)
+        self.assertEqual(filtered_feed.entries[0].title, "Article Two")
+
+    def test_filter_words_with_double_quotes(self):
+        filtered_feed_body = filter_feed(body, filtered_words='"Foo", "Bar", "One"', filtered_categories="")
+        filtered_feed = feedparser.parse(filtered_feed_body)
+
+        self.assertEqual(len(filtered_feed.entries), 1)
+        self.assertEqual(filtered_feed.entries[0].title, "Article Two")
+
     def test_filter_categories(self):
         filtered_feed_body = filter_feed(body, filtered_words="", filtered_categories="Category Four")
         filtered_feed = feedparser.parse(filtered_feed_body)
diff --git a/uv.lock b/uv.lock
