mirror of
https://github.com/navidrome/navidrome.git
synced 2026-08-01 07:21:17 +00:00
1cc0df80b1
The album, artist and playlist GetCursor built their own select and never called hydrateArtwork, so every Jellyfin list endpoint (all six stream via GetCursor) emitted entity-id image tags and no blurhash. Only GetAll hydrated, which is why Subsonic and the native API were unaffected. Each cursor now resolves its ordered/filtered/paginated id set with the cheap id-only GetAllIDs query, then streams those ids in chunks through the repo's existing GetAll, which already hydrates and applies the full select. Max/Offset are consumed by the pre-pass alone; the chunk query carries only the caller's filters, Sort and Order. This also removes a pre-existing deep-pagination cost: keeping OFFSET out of the joined query makes the pre-pass a covering index scan instead of paying the library and annotation joins for every skipped row. Benchmarked on a synthetic 100k-album DB with the real schema, page=500 at offset 90,000: 3.9ms via the id pre-pass, 52.5ms for the current shape, 192.7ms for a naive join. An unpaginated full stream costs ~24% more, which is the trade. GetAllIDs gains the annotation join whenever the caller's filters or sort reference an annotation column (same gate CountAll uses), otherwise Filters=IsFavorite and SortBy=PlayCount would fail in the pre-pass. The playlist pre-pass repeats GetAll's columns so ORDER BY keeps resolving to playlist.name rather than the joined user.name.
77 lines
2.5 KiB
Go
77 lines
2.5 KiB
Go
package persistence
|
|
|
|
import (
|
|
"context"
|
|
"iter"
|
|
"slices"
|
|
|
|
. "github.com/Masterminds/squirrel"
|
|
"github.com/navidrome/navidrome/log"
|
|
"github.com/navidrome/navidrome/model"
|
|
"github.com/pocketbase/dbx"
|
|
)
|
|
|
|
// artworkChunkSize bounds the id IN-list of each chunk fetch, keeping it under SQLite's bound
|
|
// parameter limit.
|
|
const artworkChunkSize = 500
|
|
|
|
// streamByIDs yields the rows of ids in chunks, fetching each chunk through the caller's hydrating
|
|
// fetch. Resolving ids first keeps OFFSET out of the joined query (spec §6).
|
|
func streamByIDs[S ~[]T, T any](ids []string, fetch func(chunk []string) (S, error)) iter.Seq2[T, error] {
|
|
return func(yield func(T, error) bool) {
|
|
for chunk := range slices.Chunk(ids, artworkChunkSize) {
|
|
rows, err := fetch(chunk)
|
|
if err != nil {
|
|
var zero T
|
|
yield(zero, err)
|
|
return
|
|
}
|
|
for _, row := range rows {
|
|
if !yield(row, nil) {
|
|
return
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// chunkOptions narrows the caller's options to a chunk of ids, dropping Max/Offset (the id pre-pass
|
|
// already consumed them) and reusing its Sort, which may be a seeded random expression by now.
|
|
func chunkOptions(options []model.QueryOptions, idField string) func([]string) model.QueryOptions {
|
|
var base model.QueryOptions
|
|
if len(options) > 0 {
|
|
base = model.QueryOptions{Sort: options[0].Sort, Order: options[0].Order, Filters: options[0].Filters}
|
|
}
|
|
return func(chunk []string) model.QueryOptions {
|
|
opts := base
|
|
opts.Filters = Eq{idField: chunk}
|
|
if base.Filters != nil {
|
|
opts.Filters = And{base.Filters, opts.Filters}
|
|
}
|
|
return opts
|
|
}
|
|
}
|
|
|
|
// hydrateItemImages returns per-item artwork info for a fetched page via one batched query per kind
|
|
// (never a join, see spec §6). On error it logs and returns an empty map so the page still renders.
|
|
func hydrateItemImages(ctx context.Context, db dbx.Builder, kind string, ids []string) map[string]model.ItemArtworkInfo {
|
|
if len(ids) == 0 {
|
|
return map[string]model.ItemArtworkInfo{}
|
|
}
|
|
infos, err := NewArtworkRepository(ctx, db).GetInfoForItems(kind, ids)
|
|
if err != nil {
|
|
log.Error(ctx, "Failed to hydrate artwork info onto page", "kind", kind, err)
|
|
return map[string]model.ItemArtworkInfo{}
|
|
}
|
|
return infos
|
|
}
|
|
|
|
// applyItemImage copies a hydration entry onto img; a missing entry leaves it zero (unresolved).
|
|
func applyItemImage(infos map[string]model.ItemArtworkInfo, id string, img *model.ItemImage) {
|
|
if info, ok := infos[id]; ok {
|
|
img.ImageHash = info.Hash
|
|
img.ImageAbsent = info.Absent()
|
|
img.BlurHash = info.BlurHash
|
|
}
|
|
}
|