refactor: parse category optimizations

danjac · Oct 26, 2024 · dbf35bc · dbf35bc
1 parent b533f64
commit dbf35bc
Showing 1 changed file with 14 additions and 13 deletions.
diff --git a/radiofeed/feedparser/feed_parser.py b/radiofeed/feedparser/feed_parser.py
@@ -124,6 +124,7 @@ def _handle_update(
         feed: Feed,
     ) -> None:
         categories_dct = get_categories()
+
         try:
             with transaction.atomic():
                 self._podcast_update(
@@ -134,8 +135,8 @@ def _handle_update(
                     active=not (feed.complete),
                     etag=self._parse_etag(response),
                     modified=self._parse_modified(response),
+                    keywords=self._parse_keywords(feed, categories_dct),
                     extracted_text=self._tokenize_content(feed),
-                    keywords=" ".join(self._parse_keywords(feed, categories_dct)),
                     frequency=scheduler.schedule(feed),
                     **feed.model_dump(
                         exclude={
@@ -149,7 +150,9 @@ def _handle_update(
                 self._podcast.categories.set(
                     self._parse_categories(feed, categories_dct)
                 )
+
                 self._episode_updates(feed)
+
                 self._logger.success("Feed updated")
         except DataError as exc:
             raise InvalidDataError from exc
@@ -239,21 +242,19 @@ def _podcast_update(self, **fields) -> None:
             **fields,
         )
 
-    def _parse_keywords(
-        self, feed: Feed, categories_dct: dict[str, Category]
-    ) -> Iterator[str]:
-        categories_dct = get_categories()
-
-        for value in feed.categories:
-            if value not in categories_dct:
-                yield value
+    def _parse_keywords(self, feed: Feed, categories_dct: dict[str, Category]) -> str:
+        return " ".join(
+            [value for value in feed.categories if value not in categories_dct]
+        )
 
     def _parse_categories(
         self, feed: Feed, categories_dct: dict[str, Category]
-    ) -> Iterator[Category]:
-        for value in feed.categories:
-            if value in categories_dct:
-                yield categories_dct[value]
+    ) -> list[Category]:
+        return [
+            categories_dct[value]
+            for value in feed.categories
+            if value in categories_dct
+        ]
 
     def _tokenize_content(self, feed: Feed) -> str:
         text = " ".join(