fix(scanner): optimize refresh (#4059)

* fix(artist): update RefreshStats to only process artists with recently updated media files

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: paginate Artist's RefreshStats, also replace rawSQL with Expr

Signed-off-by: Deluan <deluan@navidrome.org>

---------

Signed-off-by: Deluan <deluan@navidrome.org>

Makefile

@@ -36,8 +36,9 @@ watch: ##@Development Start Go tests in watch mode (re-run when code changes)
 	go tool ginkgo watch -tags=netgo -notify ./...
 .PHONY: watch
 
+PKG ?= ./...
 test: ##@Development Run Go tests
-	go test -tags netgo ./...
+	go test -tags netgo $(PKG)
 .PHONY: test
 
 testrace: ##@Development Run Go tests with race detector

persistence/album_repository.go

@@ -315,7 +315,7 @@ func (r *albumRepository) GetTouchedAlbums(libID int) (model.AlbumCursor, error)
 // RefreshPlayCounts updates the play count and last play date annotations for all albums, based
 // on the media files associated with them.
 func (r *albumRepository) RefreshPlayCounts() (int64, error) {
-	query := rawSQL(`
+	query := Expr(`
 with play_counts as (
     select user_id, album_id, sum(play_count) as total_play_count, max(play_date) as last_play_date
     from media_file

persistence/artist_repository.go

@@ -239,7 +239,7 @@ func (r *artistRepository) purgeEmpty() error {
 // RefreshPlayCounts updates the play count and last play date annotations for all artists, based
 // on the media files associated with them.
 func (r *artistRepository) RefreshPlayCounts() (int64, error) {
-	query := rawSQL(`
+	query := Expr(`
 with play_counts as (
     select user_id, atom as artist_id, sum(play_count) as total_play_count, max(play_date) as last_play_date
     from media_file
@@ -259,76 +259,123 @@ on conflict (user_id, item_id, item_type) do update
 	return r.executeSQL(query)
 }
 
-// RefreshStats updates the stats field for all artists, based on the media files associated with them.
+// RefreshStats updates the stats field for artists whose associated media files were updated after the oldest recorded library scan time.
-// BFR Maybe filter by "touched" artists?
+// It processes artists in batches to handle potentially large updates.
 func (r *artistRepository) RefreshStats() (int64, error) {
-	// First get all counters, one query groups by artist/role, and another with totals per artist.
+	touchedArtistsQuerySQL := `
-	// Union both queries and group by artist to get a single row of counters per artist/role.
+        SELECT DISTINCT mfa.artist_id
-	// Then format the counters in a JSON object, one key for each role.
+        FROM media_file_artists mfa
-	// Finally update the artist table with the new counters
+        JOIN media_file mf ON mfa.media_file_id = mf.id
-	// In all queries, atom is the artist ID and path is the role (or "total" for the totals)
+        WHERE mf.updated_at > (SELECT last_scan_at FROM library ORDER BY last_scan_at ASC LIMIT 1)
-	query := rawSQL(`
+        `
--- CTE to get counters for each artist, grouped by role
+
-with artist_role_counters as (
+	var allTouchedArtistIDs []string
-    -- Get counters for each artist, grouped by role
+	if err := r.db.NewQuery(touchedArtistsQuerySQL).Column(&allTouchedArtistIDs); err != nil {
-    -- (remove the index from the role: composer[0] => composer
+		return 0, fmt.Errorf("fetching touched artist IDs: %w", err)
-    select atom as artist_id,
+	}
+
+	if len(allTouchedArtistIDs) == 0 {
+		log.Debug(r.ctx, "RefreshStats: No artists to update.")
+		return 0, nil
+	}
+	log.Debug(r.ctx, "RefreshStats: Found artists to update.", "count", len(allTouchedArtistIDs))
+
+	// Template for the batch update with placeholder markers that we'll replace
+	batchUpdateStatsSQL := `
+    WITH artist_role_counters AS (
+        SELECT jt.atom AS artist_id,
                substr(
                        replace(jt.path, '$.', ''),
                        1,
-                   case when instr(replace(jt.path, '$.', ''), '[') > 0
+                       CASE WHEN instr(replace(jt.path, '$.', ''), '[') > 0
-                            then instr(replace(jt.path, '$.', ''), '[') - 1
+                                THEN instr(replace(jt.path, '$.', ''), '[') - 1
-                        else length(replace(jt.path, '$.', ''))
+                            ELSE length(replace(jt.path, '$.', ''))
-                       end
+                           END
-           ) as role,
+               ) AS role,
-           count(distinct album_id) as album_count,
+               count(DISTINCT mf.album_id) AS album_count,
-           count(mf.id) as count,
+               count(mf.id) AS count,
-           sum(size) as size
+               sum(mf.size) AS size
-    from media_file mf
+        FROM media_file mf
-             left join json_tree(participants) jt
+        JOIN json_tree(mf.participants) jt ON jt.key = 'id' AND jt.atom IS NOT NULL
-    where atom is not null and key = 'id'
+        WHERE jt.atom IN (ROLE_IDS_PLACEHOLDER) -- Will replace with actual placeholders
-    group by atom, role
+        GROUP BY jt.atom, role
     ),
-
+    artist_total_counters AS (
--- CTE to get the totals for each artist
+        SELECT mfa.artist_id,
-artist_total_counters as (
+               'total' AS role,
-	select mfa.artist_id,
+               count(DISTINCT mf.album_id) AS album_count,
-		   'total' as role,
+               count(DISTINCT mf.id) AS count,
-		   count(distinct mf.album_id) as album_count,
+               sum(mf.size) AS size
-		   count(distinct mf.id) as count,
+        FROM media_file_artists mfa
-		   sum(mf.size) as size
+        JOIN media_file mf ON mfa.media_file_id = mf.id
-	from (select artist_id, media_file_id
+        WHERE mfa.artist_id IN (TOTAL_IDS_PLACEHOLDER) -- Will replace with actual placeholders
-		  from main.media_file_artists) as mfa
+        GROUP BY mfa.artist_id
-			 join main.media_file mf on mfa.media_file_id = mf.id
-	group by mfa.artist_id
     ),
-
+    combined_counters AS (
--- CTE to combine role and total counters
+        SELECT artist_id, role, album_count, count, size FROM artist_role_counters
-combined_counters as (
+        UNION
-	select artist_id, role, album_count, count, size
+        SELECT artist_id, role, album_count, count, size FROM artist_total_counters
-	from artist_role_counters
-	union
-	select artist_id, role, album_count, count, size
-	from artist_total_counters
     ),
-
+    artist_counters AS (
--- CTE to format the counters in a JSON object
+        SELECT artist_id AS id,
-artist_counters as (
-	select artist_id as id,
                json_group_object(
                        replace(role, '"', ''),
                        json_object('a', album_count, 'm', count, 's', size)
-		   ) as counters
+               ) AS counters
-	from combined_counters
+        FROM combined_counters
-	group by artist_id
+        GROUP BY artist_id
     )
-
+    UPDATE artist
--- Update the artist table with the new counters
+    SET stats = coalesce((SELECT counters FROM artist_counters ac WHERE ac.id = artist.id), '{}'),
-update artist
-set stats = coalesce((select counters from artist_counters where artist_counters.id = artist.id), '{}'),
        updated_at = datetime(current_timestamp, 'localtime')
-where id <> ''; -- always true, to avoid warnings`)
+    WHERE artist.id IN (UPDATE_IDS_PLACEHOLDER) AND artist.id <> '';` // Will replace with actual placeholders
-	return r.executeSQL(query)
+
+	var totalRowsAffected int64 = 0
+	const batchSize = 1000
+
+	batchCounter := 0
+	for artistIDBatch := range slice.CollectChunks(slices.Values(allTouchedArtistIDs), batchSize) {
+		batchCounter++
+		log.Trace(r.ctx, "RefreshStats: Processing batch", "batchNum", batchCounter, "batchSize", len(artistIDBatch))
+
+		// Create placeholders for each ID in the IN clauses
+		placeholders := make([]string, len(artistIDBatch))
+		for i := range artistIDBatch {
+			placeholders[i] = "?"
+		}
+		// Don't add extra parentheses, the IN clause already expects them in SQL syntax
+		inClause := strings.Join(placeholders, ",")
+
+		// Replace the placeholder markers with actual SQL placeholders
+		batchSQL := strings.Replace(batchUpdateStatsSQL, "ROLE_IDS_PLACEHOLDER", inClause, 1)
+		batchSQL = strings.Replace(batchSQL, "TOTAL_IDS_PLACEHOLDER", inClause, 1)
+		batchSQL = strings.Replace(batchSQL, "UPDATE_IDS_PLACEHOLDER", inClause, 1)
+
+		// Create a single parameter array with all IDs (repeated 3 times for each IN clause)
+		// We need to repeat each ID 3 times (once for each IN clause)
+		var args []interface{}
+		for _, id := range artistIDBatch {
+			args = append(args, id) // For ROLE_IDS_PLACEHOLDER
+		}
+		for _, id := range artistIDBatch {
+			args = append(args, id) // For TOTAL_IDS_PLACEHOLDER
+		}
+		for _, id := range artistIDBatch {
+			args = append(args, id) // For UPDATE_IDS_PLACEHOLDER
+		}
+
+		// Now use Expr with the expanded SQL and all parameters
+		sqlizer := Expr(batchSQL, args...)
+
+		rowsAffected, err := r.executeSQL(sqlizer)
+		if err != nil {
+			return totalRowsAffected, fmt.Errorf("executing batch update for artist stats (batch %d): %w", batchCounter, err)
+		}
+		totalRowsAffected += rowsAffected
+	}
+
+	log.Debug(r.ctx, "RefreshStats: Successfully updated stats.", "totalArtistsProcessed", len(allTouchedArtistIDs), "totalDBRowsAffected", totalRowsAffected)
+	return totalRowsAffected, nil
 }
 
 func (r *artistRepository) Search(q string, offset int, size int, includeMissing bool) (model.Artists, error) {

persistence/helpers.go

@@ -57,14 +57,6 @@ func toCamelCase(str string) string {
 	})
 }
 
-// rawSQL is a string that will be used as is in the SQL query executor
-// It does not support arguments
-type rawSQL string
-
-func (r rawSQL) ToSql() (string, []interface{}, error) {
-	return string(r), nil, nil
-}
-
 func Exists(subTable string, cond squirrel.Sqlizer) existsCond {
 	return existsCond{subTable: subTable, cond: cond, not: false}
 }

persistence/library_repository.go

@@ -136,7 +136,7 @@ func (r *libraryRepository) ScanEnd(id int) error {
 		return err
 	}
 	// https://www.sqlite.org/pragma.html#pragma_optimize
-	_, err = r.executeSQL(rawSQL("PRAGMA optimize=0x10012;"))
+	_, err = r.executeSQL(Expr("PRAGMA optimize=0x10012;"))
 	return err
 }
 

persistence/tag_repository.go

@@ -60,7 +60,7 @@ where tag.id = updated_values.id;
 `
 	for _, table := range []string{"album", "media_file"} {
 		start := time.Now()
-		query := rawSQL(fmt.Sprintf(template, table))
+		query := Expr(fmt.Sprintf(template, table))
 		c, err := r.executeSQL(query)
 		log.Debug(r.ctx, "Updated tag counts", "table", table, "elapsed", time.Since(start), "updated", c)
 		if err != nil {