fix sync issue, frontend improvement

src/reader/sync.rs

@@ -64,6 +64,27 @@ fn create_feed_item(item: Item, feed: &Feed, connection: &mut PgConnection) {
     let item_title = item.title.clone().unwrap();
     log::info!("Create feed item: {}", item_title);
 
+    // Resolve the publication date before any HTML parsing or DB work so we can
+    // bail out early for old articles. Items without a pub_date are treated as
+    // current (inserted unconditionally) — feeds that don't publish dates are
+    // typically small/curated enough that this is fine.
+    let mut time: NaiveDateTime = Local::now().naive_local();
+    if let Some(pub_date) = item.pub_date() {
+        time = match get_date(pub_date) {
+            Ok(date) => date,
+            Err(err) => {
+                log::error!("could not parse pub date: {}", err);
+                time
+            }
+        };
+    }
+
+    let cutoff = Local::now().naive_local() - Duration::days(14);
+    if time < cutoff {
+        log::info!("Skipping item {} (older than 2 weeks).", item_title);
+        return;
+    }
+
     let base_content: &str = item.content().or(item.description()).unwrap_or_default();
 
     let frag = Html::parse_fragment(base_content);
@@ -96,17 +117,6 @@ fn create_feed_item(item: Item, feed: &Feed, connection: &mut PgConnection) {
         .unwrap();
 
     if existing_item.is_empty() {
-        log::info!("{:?}", item.pub_date());
-        let mut time: NaiveDateTime = Local::now().naive_local();
-        if item.pub_date().is_some() {
-            time = match get_date(item.pub_date().unwrap()) {
-                Ok(date) => date,
-                Err(err) => {
-                    log::error!("could not unwrap pub date: {}", err);
-                    time
-                }
-            };
-        }
         let new_feed_item = NewFeedItem::new(
             feed.id,
             content.clone(),
@@ -331,6 +341,72 @@ mod tests {
             .ok();
     }
 
+    #[actix_web::test]
+    async fn create_feed_item_skips_articles_older_than_two_weeks() {
+        let mut connection = establish_connection();
+        let suffix = unique_suffix();
+
+        let new_user = NewUser::new(
+            format!("age_skip_test_{suffix}"),
+            format!("age_skip_{suffix}@example.test"),
+            "secret".to_string(),
+        );
+        let user: User = diesel::insert_into(users::table)
+            .values(&new_user)
+            .get_result(&mut connection)
+            .unwrap();
+
+        let new_feed = NewFeed::new(
+            format!("Age skip test feed {suffix}"),
+            format!("https://example.test/feed/{suffix}"),
+            user.id,
+        );
+        let feed: Feed = diesel::insert_into(feed::table)
+            .values(&new_feed)
+            .get_result(&mut connection)
+            .unwrap();
+
+        // Item with a pub_date 20 days ago — should be ignored by create_feed_item.
+        let old_date = (Local::now() - Duration::days(20))
+            .format("%a, %d %b %Y %H:%M:%S %z")
+            .to_string();
+        let mut old_item = Item::default();
+        old_item.set_title(Some(format!("Old article {suffix}")));
+        old_item.set_link(Some(format!("https://example.test/old/{suffix}")));
+        old_item.set_pub_date(Some(old_date));
+        old_item.set_content(Some("<p>old</p>".to_string()));
+
+        // Item without a pub_date — treated as current, should be inserted.
+        let mut fresh_item = Item::default();
+        fresh_item.set_title(Some(format!("Fresh article {suffix}")));
+        fresh_item.set_link(Some(format!("https://example.test/fresh/{suffix}")));
+        fresh_item.set_content(Some("<p>fresh</p>".to_string()));
+
+        create_feed_item(old_item, &feed, &mut connection);
+        create_feed_item(fresh_item, &feed, &mut connection);
+
+        let items: Vec<FeedItem> = feed_item::table
+            .filter(feed_id.eq(feed.id))
+            .load(&mut connection)
+            .unwrap();
+
+        assert_eq!(1, items.len(), "old item should have been skipped");
+        assert!(
+            items[0].title.contains("Fresh article"),
+            "only the fresh item should be present"
+        );
+
+        diesel::delete(feed_item::table.filter(feed_id.eq(feed.id)))
+            .execute(&mut connection)
+            .ok();
+        diesel::delete(feed::table.filter(feed::id.eq(feed.id)))
+            .execute(&mut connection)
+            .ok();
+        diesel::delete(users::table.filter(users::id.eq(user.id)))
+            .execute(&mut connection)
+            .ok();
+    }
+
     #[actix_web::test]
     async fn create_feed_item_does_not_duplicate_existing_items() {
         let mut connection = establish_connection();