Fix the image cache

Author: subdavis

rss_glue/feeds/cache.py

@@ -1,27 +1,3 @@
-"""
-CacheFeed - A composable feed wrapper that caches images from post content.
-
-This module provides a CacheFeed class that wraps another feed and automatically
-downloads and caches images referenced in post HTML content. This helps avoid:
-- CORS issues when displaying images from external sources
-- Broken links from expired or unstable image URLs
-
-Usage example:
-    from rss_glue.feeds import RssFeed, CacheFeed
-
-    # Create a source feed
-    source = RssFeed("example", "https://example.com/feed.xml")
-
-    # Wrap it with CacheFeed to cache images
-    cached = CacheFeed(source)
-
-    # Use cached feed in your artifacts
-    artifacts = [HtmlOutput(cached)]
-
-The CacheFeed stores images in the file cache under an "images/" namespace
-and rewrites <img> tags to point to the locally cached versions.
-"""
-
 import re
 from dataclasses import dataclass
 from pathlib import Path
@@ -123,17 +99,6 @@ def _cache_image(self, url: str) -> Optional[Path]:
             self.logger.error(f"Failed to download image {url}: {e}")
             return None
 
-    @staticmethod
-    def load(obj: dict, source: "CacheFeed"):
-        """
-        Override ReferenceFeedItem.load() to get the subpost from source.source
-        instead of directly from source.
-        """
-        obj["subpost"] = source.source.post(obj["subpost"])
-        if not obj["subpost"]:
-            source.logger.error(f"missing reference ns={obj['namespace']} subpost={obj['id']}")
-        return obj
-
 
 class CacheFeed(feed.ReferenceFeed):
     """

rss_glue/feeds/feed.py

@@ -6,6 +6,7 @@
 
 from rss_glue.logger import logger
 from rss_glue.resources import global_config, utc_now
+from rss_glue.utils import from_dict
 
 
 @dataclass
@@ -54,7 +55,8 @@ def to_dict(self) -> dict:
 @dataclass
 class ReferenceFeedItem(FeedItem):
     """
-    A ReferenceFeedItem is a reference to another feed item.
+    A ReferenceFeedItem contains a reference to another feed item.
+    It is only useful if your feed needs to store ADDITIONAL data to the post.
     """
 
     subpost: FeedItem
@@ -72,6 +74,11 @@ def hashkey(self):
 
     @staticmethod
     def load(obj: dict, source: "BaseFeed"):
+        """
+        A reference feed has a load function which should take the contents of the
+        raw JSON object kept in the cache and return a fully "hydrated" FeedItem
+        which probably references other feed item(s).
+        """
         obj["subpost"] = source.post(obj["subpost"])
         if not obj["subpost"]:
             logger.error(f"missing reference ns={obj['namespace']} subpost={obj['id']}")
@@ -276,9 +283,7 @@ def post(self, post_id: str) -> Optional[FeedItem]:
         if not cached:
             return None
 
-        # Load the cached post
-        loaded = self.post_cls.load(cached, self)
-        if not loaded:
-            return None
+        cached["subpost"] = cached["id"]
+        loaded = self.post_cls.load(cached, self.source)
 
-        return self.post_cls(**loaded)
+        return from_dict(self.post_cls, loaded)

rss_glue/utils.py

@@ -1,4 +1,5 @@
-from typing import Iterable, Optional
+from typing import Iterable, Optional, TypeVar
+import inspect
 
 import requests
 from requests.adapters import HTTPAdapter
@@ -144,3 +145,10 @@ def make_browser_session(
     session.mount("https://", adapter)
 
     return session
+
+
+T = TypeVar("T")
+
+
+def from_dict(cls: type[T], obj: dict) -> T:
+    return cls(**{k: v for k, v in obj.items() if k in inspect.signature(cls).parameters})