Add related articles under web issue articles and keep embeddings forever

Each pick's web page now lists up to five related articles: the nearest
embedded articles across the whole archive (cosine >= 0.55), excluding the
rest of the issue, excerpt-only articles and same-title copies. They are
computed once per issue after publishing and stored in issue_related
(migration 0014), since the full embedding scan is too slow per page view;
`features related [--date]` and the related-refresh job backfill older issues.

Related articles that ran in an issue link to that issue page; the rest link
to a new signed-in /articles/{id} page rendering the stored body with no
summary. /articles/{id} for a published article redirects to its issue page.

embedding_retention_days = 0 now means keep every vector (previously it would
have deleted every unrated, unpublished embedding) and is the new default, so
related lists can reach the whole archive.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AeJmv3WED4w8S3GV7mMjS7

src/web/issue.rs

@@ -960,6 +960,36 @@ struct ArticleTemplate {
     previous: Option<ArticleLink>,
     next: Option<ArticleLink>,
     issue_href: String,
+    related: Vec<RelatedView>,
+}
+
+/// An archive article that never ran in an issue, reached from a related list:
+/// the stored body with no summary, rating or issue navigation.
+#[derive(Template)]
+#[template(path = "article_standalone.html")]
+struct StandaloneArticleTemplate {
+    page: Page,
+    title: String,
+    source_url: String,
+    byline: Option<String>,
+    source: Source,
+    meta_line: String,
+    social_line: Option<String>,
+    excerpt_only: bool,
+    body_html: String,
+    read_online_url: String,
+    dashboard_href: String,
+}
+
+/// One entry of the "Related" list under an issue article.
+#[derive(Debug)]
+struct RelatedView {
+    title: String,
+    href: String,
+    author: Option<String>,
+    source: Source,
+    reading_minutes: i64,
+    date: String,
 }
 
 #[derive(Template)]
@@ -1114,6 +1144,7 @@ pub async fn article(
     let is_admin = viewer.role == crate::web::users::Role::Admin;
     let config = state.config();
     let active_nav = if view.is_latest { "latest" } else { "archive" };
+    let related = related_views(&state.db, &config, is_admin, date, article_id).await?;
     let mut page = Page::new(article.title.clone(), Some(viewer.clone()), active_nav);
     page.flash = take_flash(&session).await?;
     Ok(Html(ArticleTemplate {
@@ -1150,10 +1181,104 @@ pub async fn article(
         previous,
         next,
         issue_href: issue_href(date),
+        related,
     })
     .into_response())
 }
 
+/// `/articles/{id}`: any stored article on its own page. One that ran in an
+/// issue redirects to its (latest) issue page, which carries the summary and
+/// related list.
+pub async fn standalone(
+    State(state): State<AppState>,
+    auth: AuthSession,
+    Extension(session): Extension<Session>,
+    Path(article_id): Path<ArticleId>,
+) -> Result<Response, WebError> {
+    let viewer = auth
+        .user()
+        .await
+        .map(Viewer::from)
+        .ok_or_else(|| WebError::Unauthenticated {
+            next: standalone_href(article_id),
+        })?;
+    let Some(article) = state.db.get_article(article_id).await? else {
+        return Err(WebError::NotFound);
+    };
+    if let Some(date) = crate::related::latest_issues(&state.db, &[article_id])
+        .await?
+        .remove(&article_id)
+    {
+        return Ok(Redirect::to(&article_href(date, article_id)).into_response());
+    }
+    let is_admin = viewer.role == crate::web::users::Role::Admin;
+    let config = state.config();
+    let mut page = Page::new(article.title.clone(), Some(viewer), "archive");
+    page.flash = take_flash(&session).await?;
+    Ok(Html(StandaloneArticleTemplate {
+        page,
+        title: article.title.clone(),
+        source_url: article.canonical_url.clone(),
+        byline: article.author.as_ref().map(|author| format!("By {author}")),
+        source: Source::new(&article, &config, is_admin),
+        meta_line: format!(
+            "{} words · ~{} min read",
+            thousands(article.word_count),
+            article.reading_minutes()
+        ),
+        social_line: chapters::social_line(&article.social),
+        excerpt_only: article.excerpt_only,
+        body_html: prepare_body(&article.content_html),
+        read_online_url: article.url.clone(),
+        dashboard_href: format!("/dashboard/articles/{}", article.id),
+    })
+    .into_response())
+}
+
+/// The stored related links of one pick, each pointing at the issue page it
+/// ran on or, failing that, its standalone page.
+async fn related_views(
+    db: &Db,
+    config: &Config,
+    is_admin: bool,
+    date: Date,
+    article_id: ArticleId,
+) -> anyhow::Result<Vec<RelatedView>> {
+    let links = crate::related::for_article(db, date, article_id).await?;
+    if links.is_empty() {
+        return Ok(Vec::new());
+    }
+    let ids: Vec<ArticleId> = links.iter().map(|link| link.article_id).collect();
+    let articles = db.get_articles(&ids).await?;
+    let issues = crate::related::latest_issues(db, &ids).await?;
+    Ok(links
+        .iter()
+        .filter_map(|link| {
+            let article = articles.get(&link.article_id)?;
+            let href = match issues.get(&link.article_id) {
+                Some(date) => article_href(*date, article.id),
+                None => standalone_href(article.id),
+            };
+            Some(RelatedView {
+                title: article.title.clone(),
+                href,
+                author: article.author.clone(),
+                source: Source::new(article, config, is_admin),
+                reading_minutes: article.reading_minutes(),
+                date: short_date(article.published_at.unwrap_or(article.first_seen), config),
+            })
+        })
+        .collect())
+}
+
+/// "Aug 15, 2026" in the configured zone.
+fn short_date(timestamp: jiff::Timestamp, config: &Config) -> String {
+    match config.tz() {
+        Ok(tz) => timestamp.to_zoned(tz).strftime("%b %-d, %Y").to_string(),
+        Err(_) => timestamp.strftime("%b %-d, %Y").to_string(),
+    }
+}
+
 pub async fn world(
     State(state): State<AppState>,
     auth: AuthSession,
@@ -1335,6 +1460,10 @@ fn article_href(date: Date, article_id: ArticleId) -> String {
     format!("/issues/{date}/articles/{article_id}")
 }
 
+fn standalone_href(article_id: ArticleId) -> String {
+    format!("/articles/{article_id}")
+}
+
 fn issue_href(date: Date) -> String {
     format!("/issues/{date}")
 }
@@ -2198,6 +2327,98 @@ mod tests {
         assert!(!reader.contains(&feed_href));
     }
 
+    #[tokio::test]
+    async fn related_links_reach_issue_pages_and_standalone_archive_articles() {
+        let (_dir, db, source) = seeded_issue(true).await;
+        crate::web::users::add(&db, "reader", "correct horse battery", false)
+            .await
+            .unwrap();
+        let pick = &source.lineup.picks[0].article;
+        let sibling = &source.lineup.picks[1].article;
+        let mut archived = pick.clone();
+        archived.id = 0;
+        archived.canonical_url = "https://example.com/archived-story".into();
+        archived.title = "An archived story nobody picked".into();
+        archived.content_html = "<p>Archived body text.</p>".into();
+        let archived_id = db.upsert_article(&archived).await.unwrap();
+        for (position, related_id) in [archived_id, sibling.id].into_iter().enumerate() {
+            sqlx::query(
+                "INSERT INTO issue_related (issue_date, article_id, position, related_id, cosine)
+                 VALUES (?, ?, ?, ?, 0.8)",
+            )
+            .bind(source.meta.date.to_string())
+            .bind(pick.id)
+            .bind(position as i64)
+            .bind(related_id)
+            .execute(db.pool())
+            .await
+            .unwrap();
+        }
+        let app = crate::server::router(crate::server::AppState::new(
+            db,
+            crate::config::Config::default(),
+            None,
+        ));
+        let cookie = login_cookie(&app, "reader", "correct horse battery").await;
+        let get = |uri: String, cookie: Option<String>| {
+            let app = app.clone();
+            async move {
+                let mut request = Request::builder().uri(uri);
+                if let Some(cookie) = cookie {
+                    request = request.header(header::COOKIE, cookie);
+                }
+                app.oneshot(request.body(Body::empty()).unwrap())
+                    .await
+                    .unwrap()
+            }
+        };
+
+        let page = response_text(
+            get(
+                format!("/issues/{}/articles/{}", source.meta.date, pick.id),
+                Some(cookie.clone()),
+            )
+            .await,
+        )
+        .await;
+        let related = page
+            .split_once("id=\"related-heading\"")
+            .expect("related list")
+            .1;
+        let archived_at = related
+            .find(&format!("href=\"/articles/{archived_id}\""))
+            .expect("unpicked article links to its standalone page");
+        let sibling_at = related
+            .find(&format!(
+                "href=\"/issues/{}/articles/{}\"",
+                source.meta.date, sibling.id
+            ))
+            .expect("a picked article links to its issue page");
+        assert!(archived_at < sibling_at, "stored order is kept");
+
+        let standalone = get(format!("/articles/{archived_id}"), Some(cookie.clone())).await;
+        assert_eq!(standalone.status(), StatusCode::OK);
+        let standalone = response_text(standalone).await;
+        assert!(standalone.contains("An archived story nobody picked"));
+        assert!(standalone.contains("Archived body text."));
+
+        let picked = get(format!("/articles/{}", pick.id), Some(cookie)).await;
+        assert_eq!(picked.status(), StatusCode::SEE_OTHER);
+        assert_eq!(
+            picked.headers()[header::LOCATION],
+            format!("/issues/{}/articles/{}", source.meta.date, pick.id)
+        );
+
+        let anonymous = get(format!("/articles/{archived_id}"), None).await;
+        assert!(anonymous.status().is_redirection());
+        assert!(
+            anonymous.headers()[header::LOCATION]
+                .to_str()
+                .unwrap()
+                .starts_with("/login")
+        );
+    }
+
     #[tokio::test]
     async fn fallback_full_issue_recovers_colophon_world_and_behind() {
         let (_dir, db, source) = seeded_issue(false).await;