Page the version name sweep with a resume cursor (#194)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

src/database/version_dao.rs

@@ -230,12 +230,15 @@ pub trait VersionDao: Sync + Send {
 
     /// Rows in a library whose file name follows the version convention
     /// (`<name>.<kind>[.<n>].<ext>`), have a hash and aren't linked yet.
-    /// Returns `(rel_path, content_hash)`, oldest first.
+    /// Returns `(id, rel_path, content_hash)` ordered by row id, starting
+    /// after `after_id` (the caller's resume cursor, so rows that never
+    /// resolve can't starve the ones behind them).
     fn unlinked_name_candidates(
         &mut self,
         library_id: i32,
+        after_id: i32,
         limit: i64,
-    ) -> Result<Vec<(String, String)>, DbError>;
+    ) -> Result<Vec<(i32, String, String)>, DbError>;
 
     /// Copy the original's date, GPS and camera fields onto every version
     /// row whose copy is stale. Returns the number of rows updated.
@@ -723,11 +726,14 @@ impl VersionDao for SqliteVersionDao {
     fn unlinked_name_candidates(
         &mut self,
         library_id: i32,
+        after_id: i32,
         limit: i64,
-    ) -> Result<Vec<(String, String)>, DbError> {
+    ) -> Result<Vec<(i32, String, String)>, DbError> {
         trace_db_call("query", "unlinked_name_candidates", |_span| {
             #[derive(QueryableByName)]
             struct Row {
+                #[diesel(sql_type = Integer)]
+                id: i32,
                 #[diesel(sql_type = Text)]
                 rel_path: String,
                 #[diesel(sql_type = Text)]
@@ -739,8 +745,8 @@ impl VersionDao for SqliteVersionDao {
                 .collect::<Vec<_>>()
                 .join(" OR ");
             let sql = format!(
-                "SELECT rel_path, content_hash FROM image_exif \
-                 WHERE library_id = ? AND content_hash IS NOT NULL AND ({kinds}) \
+                "SELECT id, rel_path, content_hash FROM image_exif \
+                 WHERE library_id = ? AND id > ? AND content_hash IS NOT NULL AND ({kinds}) \
                    AND NOT EXISTS (SELECT 1 FROM photo_versions pv \
                                     WHERE pv.version_hash = image_exif.content_hash) \
                  ORDER BY id LIMIT ?"
@@ -748,11 +754,12 @@ impl VersionDao for SqliteVersionDao {
             let mut conn = self.connection.lock().expect("Unable to get VersionDao");
             sql_query(sql)
                 .bind::<Integer, _>(library_id)
+                .bind::<Integer, _>(after_id)
                 .bind::<BigInt, _>(limit)
                 .load::<Row>(&mut *conn)
                 .map(|rows| {
                     rows.into_iter()
-                        .map(|r| (r.rel_path, r.content_hash))
+                        .map(|r| (r.id, r.rel_path, r.content_hash))
                         .collect()
                 })
                 .map_err(|e| anyhow::anyhow!("Query error: {}", e))
@@ -1089,8 +1096,18 @@ mod tests {
         insert_exif(&c, 2, "a/y.jpg.restored.png", "h4");
         dao.link_version(&link("h2", "ox", "a/x.jpg.edit.2.png"))
             .unwrap_or(LinkOutcome::Linked);
-        let got = dao.unlinked_name_candidates(1, 100).unwrap();
-        assert_eq!(got, vec![("a/x.jpg.restored.png".to_string(), "h1".into())]);
+        let got = dao.unlinked_name_candidates(1, 0, 100).unwrap();
+        assert_eq!(got.len(), 1);
+        assert_eq!(
+            (got[0].1.as_str(), got[0].2.as_str()),
+            ("a/x.jpg.restored.png", "h1")
+        );
+        // The cursor skips rows at or before it.
+        assert!(
+            dao.unlinked_name_candidates(1, got[0].0, 100)
+                .unwrap()
+                .is_empty()
+        );
     }
 
     #[test]

src/versions/integration_tests.rs

@@ -253,6 +253,46 @@ async fn name_link_pairs_raw_and_jpeg_with_the_right_original() {
     );
 }
 
+#[actix_rt::test]
+async fn unresolvable_name_candidates_do_not_starve_later_ones() {
+    let fx = fx();
+    // Many matching rows with no original anywhere, then one linkable pair.
+    for i in 0..120 {
+        fx.put(
+            &format!("album_a/aiedits/orphan{i}.jpg.edit.png"),
+            format!("orphan {i}").as_bytes(),
+        );
+    }
+    let original = fx.put("album_b/real.jpg", b"real original");
+    let version = fx.put("album_b/aiedits/real.jpg.restored.png", b"real version");
+
+    let exif = fx.exif_arc();
+    let mut cursor = 0;
+    let mut total = 0;
+    // One 50-row page per call: the old "ORDER BY id LIMIT n" would return
+    // the same 50 orphans forever.
+    for _ in 0..5 {
+        total += sweep_name_links_from(
+            &fx.library,
+            &exif,
+            &fx.state.version_dao,
+            &mut cursor,
+            50,
+            1,
+        );
+    }
+    assert_eq!(total, 1);
+    let link = fx
+        .state
+        .version_dao
+        .lock()
+        .unwrap()
+        .get_link(&version)
+        .unwrap()
+        .unwrap();
+    assert_eq!(link.original_hash, original);
+}
+
 #[actix_rt::test]
 async fn name_link_waits_for_the_original_to_be_hashed() {
     let fx = fx();

src/versions/mod.rs

@@ -266,26 +266,83 @@ pub fn sweep_name_links(
     exif_dao: &Arc<Mutex<Box<dyn ExifDao>>>,
     version_dao: &Arc<Mutex<Box<dyn VersionDao>>>,
 ) -> usize {
-    let candidates = match version_dao
+    let mut cursor = name_sweep_cursors()
         .lock()
-        .expect("Unable to lock VersionDao")
-        .unlinked_name_candidates(library.id, NAME_SWEEP_LIMIT)
-    {
-        Ok(c) => c,
-        Err(e) => {
-            warn!(
-                "versions: name sweep failed for '{}': {:?}",
-                library.name, e
-            );
-            return 0;
-        }
-    };
+        .expect("name sweep cursors")
+        .get(&library.id)
+        .copied()
+        .unwrap_or(0);
+    let linked = sweep_name_links_from(
+        library,
+        exif_dao,
+        version_dao,
+        &mut cursor,
+        NAME_SWEEP_LIMIT,
+        NAME_SWEEP_PAGES,
+    );
+    name_sweep_cursors()
+        .lock()
+        .expect("name sweep cursors")
+        .insert(library.id, cursor);
+    linked
+}
+
+/// Pages examined per sweep call.
+const NAME_SWEEP_PAGES: usize = 4;
+
+/// Per-library resume point (last candidate row id visited) for the name
+/// sweep. Candidates that never resolve (no original in the library) stay
+/// matching forever; paging past them keeps them from starving the rows
+/// behind them. In memory only: a restart just rescans from the start.
+fn name_sweep_cursors() -> &'static Mutex<std::collections::HashMap<i32, i32>> {
+    static CURSORS: std::sync::OnceLock<Mutex<std::collections::HashMap<i32, i32>>> =
+        std::sync::OnceLock::new();
+    CURSORS.get_or_init(Default::default)
+}
+
+/// One sweep: up to `max_pages` pages of `page` candidates from `cursor`,
+/// advancing it and wrapping to 0 past the last candidate (then continuing
+/// from the start, so a call never ends having examined nothing).
+fn sweep_name_links_from(
+    library: &Library,
+    exif_dao: &Arc<Mutex<Box<dyn ExifDao>>>,
+    version_dao: &Arc<Mutex<Box<dyn VersionDao>>>,
+    cursor: &mut i32,
+    page: i64,
+    max_pages: usize,
+) -> usize {
     let mut linked = 0;
-    for (rel, hash) in candidates {
-        if let Some(link) = name_link(library, &rel, &hash, exif_dao)
-            && record_link(version_dao, &link)
+    let mut wrapped = *cursor == 0;
+    for _ in 0..max_pages {
+        let candidates = match version_dao
+            .lock()
+            .expect("Unable to lock VersionDao")
+            .unlinked_name_candidates(library.id, *cursor, page)
         {
-            linked += 1;
+            Ok(c) => c,
+            Err(e) => {
+                warn!(
+                    "versions: name sweep failed for '{}': {:?}",
+                    library.name, e
+                );
+                return linked;
+            }
+        };
+        let short = (candidates.len() as i64) < page;
+        for (id, rel, hash) in candidates {
+            *cursor = id;
+            if let Some(link) = name_link(library, &rel, &hash, exif_dao)
+                && record_link(version_dao, &link)
+            {
+                linked += 1;
+            }
+        }
+        if short {
+            *cursor = 0;
+            if wrapped {
+                break;
+            }
+            wrapped = true;
         }
     }
     linked