|
|
@@ -44,6 +44,7 @@ async def refresh_clusters(topic: str | None = None, limit: int = 80) -> None:
|
|
|
per_feed[feed_url].append(article)
|
|
|
|
|
|
changed_articles: list[dict[str, Any]] = []
|
|
|
+ changed_feed_urls: list[str] = []
|
|
|
for feed_url, feed_articles in per_feed.items():
|
|
|
material = "\n".join(
|
|
|
f"{a.get('title','')}|{a.get('url','')}|{a.get('timestamp','')}"
|
|
|
@@ -56,7 +57,7 @@ async def refresh_clusters(topic: str | None = None, limit: int = 80) -> None:
|
|
|
logger.info("refresh unchanged feed_url=%s count=%s topic=%s", feed_url, len(feed_articles), topic)
|
|
|
else:
|
|
|
logger.info("refresh changed feed_url=%s count=%s topic=%s", feed_url, len(feed_articles), topic)
|
|
|
- store.set_feed_state(feed_key, last_hash, len(feed_articles))
|
|
|
+ changed_feed_urls.append(feed_url)
|
|
|
changed_articles.extend(feed_articles)
|
|
|
|
|
|
if not changed_articles:
|
|
|
@@ -151,5 +152,13 @@ async def refresh_clusters(topic: str | None = None, limit: int = 80) -> None:
|
|
|
retention_days=NEWS_RETENTION_DAYS,
|
|
|
interval_hours=NEWS_PRUNE_INTERVAL_HOURS,
|
|
|
)
|
|
|
+ for feed_url in changed_feed_urls:
|
|
|
+ feed_articles = per_feed[feed_url]
|
|
|
+ material = "\n".join(
|
|
|
+ f"{a.get('title','')}|{a.get('url','')}|{a.get('timestamp','')}"
|
|
|
+ for a in feed_articles
|
|
|
+ )
|
|
|
+ last_hash = hashlib.sha1(material.encode("utf-8")).hexdigest()
|
|
|
+ store.set_feed_state(feed_url, last_hash, len(feed_articles))
|
|
|
store.set_meta("last_refresh_at", datetime.now(timezone.utc).isoformat())
|
|
|
logger.info("refresh prune_result=%s", prune_result)
|
