navidrome/core/artwork/processor.go
Deluan Quintão 59448e9283
fix(scrobbler): back off when a provider asks us to, instead of retrying per play (#6028)
* feat(agents): retry-later error type with optional server delay

Add agents.ErrRetryLater and agents.RetryLaterError, which carries the
delay requested by an external service (e.g. ListenBrainz's
X-RateLimit-Reset-In). scrobbler.ErrRetryLater becomes an alias of the new
sentinel, so existing errors.Is checks and the plugin error-string protocol
keep working unchanged. Groundwork for honoring server-requested retry
delays across scrobbling, metadata agents and artwork.

Song.Equals tests moved to song_test.go to enable external test package.

* fix(scrobbler): honor backoff window and server-requested retry delay

ListenBrainz 429s were decoded into a typed error that classified as
unrecoverable, silently discarding the scrobble (a JSON-bodied 429 was
measured live). The client now maps any 429 to agents.RetryLaterError,
carrying X-RateLimit-Reset-In when present (capped at 1h). Last.fm error 29
(rate limit) is now retryable like 11/16. The buffer's drain loop no longer
lets wake signals bypass an active backoff window - new plays enqueue but
drain only when the window closes - and the wait honors the server delay
via max(backoff, retryIn).

* feat(agents): skip cooling-down agents in aggregate calls

When an agent reports retry-later, remember a per-agent cooldown deadline
(the server-requested delay, or 1 minute when unspecified) and skip that
agent in all aggregate metadata calls until it passes. A round that found
no data but skipped or saw a throttled agent returns ErrRetryLater instead
of ErrNotFound, so callers cannot mistake rate limiting for a definitive
'no data' answer.

* feat(artwork): honor server-requested retry delay when rescheduling

When an external image lookup fails with a retry-later error carrying a
delay (e.g. a 429 with X-RateLimit-Reset-In), the chain trace carries the
largest such hint back to the worker, which reschedules the item at
max(exponential backoff, server delay) instead of backoff alone.

* feat(plugins): retry-later with optional delay for scrobbler and agent plugins

Scrobbler plugins can now return scrobbler(retry_later:N) to request a
retry in N seconds (capped at 1h); the bare token keeps its old meaning.
Metadata-agent plugins, which had no error vocabulary at all, gain the
parallel agent(retry_later[:N]) token, mapped to agents.RetryLaterError so
the aggregate's cooldown and the artwork worker honor plugin throttling
the same way as built-in agents.

* fix: address whole-branch review findings for retry-later handling

Narrow the aggregate's throttled rule to the spec sentence: core.Agents returns
ErrRetryLater only when no agent answered at all (all skipped-cooling or
retry-later). An agent that does not implement the called method now returns an
internal errUnsupported instead of ErrNotFound, so it counts as "did not run" —
without that, the always-appended local agent would answer for biography, URL
and images and make ErrRetryLater unreachable.

Wire the consequence in core/external: a throttled round no longer stamps
ExternalInfoUpdatedAt (artist and album), so the empty result is not cached for
the TTL, and TopSongs maps ErrRetryLater to the same empty-200 the not-found
path already produced instead of a new client-facing error.

Move the Last.fm code-29 mapping into the client's central error construction so
every metadata path produces RetryLaterError, and map ListenBrainz's body-level
code 429 (sent with a non-429 HTTP status) the same way.

Clamp server- and plugin-requested delays in seconds before scaling to a
Duration, in all three parse sites: a header of 18446744074 wrapped past 2^64 and
came out as a 0.29s delay.

Also: extract the artwork worker's reschedule computation into retryDelay() and
cover both it and the trace RetryIn wiring with tests; collapse the double regex
call in mapScrobblerError; drop capabilities.ScrobblerErrorRetryLaterIn (ndpgen
never emits funcs, so plugin authors could not reach it); regenerate the PDKs so
MetadataAgentError reaches the Go and Rust SDKs; de-flake the cooldown tests
(long RetryIn for the skip case, separate expiry spec); and cover the max()
retry-delay aggregation across users in the scrobble buffer.

* refactor: dedupe retry-later parsing and simplify error collection

- Add agents.NewRetryLater and agents.RetryLaterFromSeconds, with a single
  1h cap, replacing the parse+clamp+multiply logic and the maxRetryInSeconds
  constant duplicated across listenbrainz, plugins and the agent adapter.
- Move HTTP header parsing to httpclient.RetryAfter, so the transport layer
  owns it and stays domain-agnostic; drop retryInFromHeaders from the
  ListenBrainz client. Covered by a new Ginkgo table in that package.
- Collapse the two near-identical plugin retry_later regexes into one
  parseRetryLater(prefix, msg) shared by the agent and scrobbler adapters.
- Fold the duplicated noteRetryIn snippet from fetchArtistImage and
  fetchAlbumImage into recordAgent, which already branched on the same
  isTransientExternal condition.
- Replace the atomic.Bool + note() closure in populateArtistInfo with
  errgroup's own error collection; the group carries no context, so a
  returned error does not cancel its siblings.
- Reuse recoveringScrobbler for the per-user delay test instead of a third
  double, and switch fakeScrobbler's mutex-guarded error to the
  atomic.Pointer idiom already used in the same package.

* refactor(listenbrainz): keep rate-limit header parsing in the adapter

The X-RateLimit-Reset-In header is ListenBrainz's own convention, not a
shared one: Last.fm sends no rate-limit headers at all and reports its
limit as a body code, and no other integration in tree sends Retry-After.
A parser in utils/httpclient implied a uniformity across services that
does not exist, so it moves back next to the only client that can know
which header its service sends.

* refactor(agents): collapse the retry-later sentinel and error into one type

ErrRetryLater is now the zero-delay RetryLaterError rather than a separate
errors.New value, so errors.Is and errors.AsType both match the sentinel and
every delay-carrying variant. That removes the trap where a bare sentinel
silently skipped the AsType path, and lets every consumer read the delay off
the error directly: the RetryIn accessor and the two constructors are gone,
with the policy cap applied where untrusted input is parsed.

* refactor(agents): split the cooldown store from the per-dispatch tally

The cooldown map and mutex become a cooldowns value with active/park, holding
no knowledge of errors; agentAttempts records one dispatch's outcomes and owns
the classification that noteAgentError used to hide behind a bool. The three
dispatch loops now touch a single object: skip folds the cooldown check and the
throttled flag into one call, so the store never appears in the loops.

* refactor(agents): share one dispatch loop between the agent call helpers

callAgentMethod and callAgentSliceMethod ran identical loops, differing only in
how they test a result for emptiness: a slice cannot be compared against its
zero value, so the two could not share a constraint. Both now delegate to
callAgent, which takes that test as a parameter. Keeping the loop in one place
matters more than the lines saved: it holds the cooldown skip, the attempt
recording and the empty-dispatch verdict, and a fix applied to one copy but not
the other would be silent.

* test: cover the two retry-later paths a mutation could break silently

Both gaps were proven, not guessed: making the artwork worker pass 0 instead
of the collected hint left all 386 specs green, and replacing the default
agent cooldown with 0 left the agents suite green. The worker test drives a
throttled image agent through drain and asserts the persisted retry_at, and
the cooldown test parks an agent that asked to be retried without naming a
delay, which is what Last.fm does on every rate limit.

* refactor(artwork): carry the external failure as an error, not a flag plus a trace field

The retry delay was riding on ChainTrace, a diagnostic that gets persisted, while
the very same signal — an external source faulted — already travelled by value as
resolution.extError. That was two mechanisms for one idea, and it put control-flow
state inside a serializable trace.

resolution.extError and chainState.extErr become the error itself, so a caller
checks err != nil for the fault and errors.AsType for the delay the provider asked
for. The agent loops return that error last, per convention, and longerRetry keeps
whichever failure wants the longer wait. ChainTrace goes back to holding only steps
and no longer imports core/agents.

* fix(artwork): check the resolve error before reading its resolution

Reading res.extError before the err check was safe only because every error path
in resolve returns a bare resolution{}; a future path returning a partly-filled
one would have been read silently. The failure path now returns no delay
explicitly.

* test(artwork): assert the delay acquire reports, not just its downstream effect

acquire's retry delay was only covered through the worker's persisted retry_at,
one layer away from where the value is computed. Both outcomes are now pinned at
the processor: a plain failure asks for nothing, a throttled provider's delay is
passed through.

* refactor: share the retry-seconds parse and drop the backoff deadline arithmetic

The clamp-before-scaling invariant lived in two parsers and was independently
re-tested in three files with the same magic number; a fix applied to one copy
would have left the others wrapping a huge value down to a fraction of a second.
It moves to agents.ParseRetryIn.

The buffer tracked an absolute retryDeadline only to re-arm a timer that was
already armed for the same instant; a backingOff flag says the same thing without
the arithmetic. The plugin token regex now carries its capability in the pattern
instead of capturing and comparing, so another capability's token in the same
message cannot mask it. resolution.extError becomes extErr, matching its
chainState counterpart.

* fix(agents): keep the longer cooldown when parks overlap

Calls to one agent overlap, so a short cooldown could land after a long one
started and cut it short. park now keeps whichever deadline is later, matching
the rule longerRetry already applies on the artwork side. No in-tree provider
can currently produce two different delays for the same agent, so this is
hardening rather than a fix for observed behaviour.

* fix(agents): parse the retry delay at a fixed width

strconv.Atoi parses into the native int, so on the 32-bit targets we ship
(linux/386, windows/386, three ARM variants) a delay above MaxInt32 seconds
overflowed and became unspecified instead of being capped. No provider sends a
68-year delay, so this is not user-visible, but the overflow tests asserted the
cap and would have failed on those architectures, where tests never run.

* fix(plugins): anchor the retry_later regex to a word boundary

Prevents a superstring like useragent(retry_later) from matching the
agent capability token.
2026-08-29 17:28:29 -04:00

357 lines
12 KiB
Go

package artwork
import (
"bytes"
"cmp"
"context"
"encoding/base64"
"errors"
"fmt"
"image"
"image/draw"
_ "image/gif" // the only artwork format with no other importer in this package
"io"
"sync"
"time"
"github.com/navidrome/navidrome/conf"
"github.com/navidrome/navidrome/consts"
"github.com/navidrome/navidrome/core/agents"
"github.com/navidrome/navidrome/core/artwork/blurhash"
"github.com/navidrome/navidrome/core/artwork/dominant"
"github.com/navidrome/navidrome/core/artwork/thumbhash"
"github.com/navidrome/navidrome/log"
"github.com/navidrome/navidrome/model"
xdraw "golang.org/x/image/draw"
)
// outcome tells the worker whether to delete the queue row (found/absent) or reschedule it.
type outcome int
const (
outcomeFound outcome = iota
// outcomeFoundStale: state was written and is served, but a higher-priority external
// source failed, so the row must retry to give it another chance.
outcomeFoundStale
outcomeAbsent
outcomeFailed
)
func (o outcome) String() string {
switch o {
case outcomeFound:
return "found"
case outcomeFoundStale:
return "foundStale"
case outcomeAbsent:
return "absent"
default:
return "failed"
}
}
// thumbnailSize is the max dimension fed to both hash encoders; thumbhash rejects anything larger.
const thumbnailSize = 100
// maxImageBytes caps a resolved image read: a user-editable ExternalImageURL could point at
// an arbitrarily large endpoint.
func maxImageBytes() int64 {
return parseSize(conf.Server.MaxImageSize, consts.DefaultMaxImageSize)
}
// maxImagePixels guards against decompression bombs: a tiny file can declare a canvas that
// image.Decode would expand into gigabytes.
const maxImagePixels = 64 << 20
// acquired is what a successful acquire persisted, handed back so the caller can warm the resize
// cache without re-reading the rows and the file it just wrote.
type acquired struct {
ia *model.ItemArtwork
mime string
data []byte
}
// processor turns one queue item into stored artwork; settling the queue row is the Worker's job.
type processor struct {
ds model.DataStore
store *ImageStore
resolver *resolver
pruneLock sync.Locker
}
// acquire resolves one queue item end to end: find an image, hash/decode/
// blurhash it, place its bytes, and persist the resulting state.
func (p *processor) acquire(ctx context.Context, item model.ArtworkQueueItem) (out outcome, got *acquired, retryIn time.Duration) {
repo := p.ds.Artwork(ctx)
start := time.Now()
defer func() {
log.Debug(ctx, "Artwork: Acquisition finished", "kind", item.ItemKind, "id", item.ItemID,
"outcome", out, "elapsed", time.Since(start))
}()
res, err := p.resolver.resolve(ctx, item)
if err != nil {
traceStage(ctx, "resolve", err)
log.Warn(ctx, "Artwork: Could not resolve item", "kind", item.ItemKind, "id", item.ItemID, err)
return outcomeFailed, nil, 0
}
if retry, ok := errors.AsType[*agents.RetryLaterError](res.extErr); ok {
retryIn = retry.RetryIn
}
if res.reader == nil {
if res.extErr != nil || res.localError {
// A fault is not a definitive "no image": never settle absent, keep serving old state.
// A chainless resolver (playlist/radio) records no step, so leave a fallback or explain is blank.
if t := traceFrom(ctx); len(t.Steps()) == 0 {
outcome := OutcomeError
if res.localError {
outcome = OutcomeUnreadable
}
t.add(TraceStep{Candidate: cmp.Or(res.source, "source"), Outcome: outcome})
}
log.Debug(ctx, "Artwork: No image, but a source faulted; keeping previous state",
"kind", item.ItemKind, "id", item.ItemID, "extErr", res.extErr, "localError", res.localError)
return outcomeFailed, nil, retryIn
}
return writeAbsent(ctx, repo, item), nil, 0
}
defer res.reader.Close()
readStart := time.Now()
data, err := readCapped(res.reader)
if err != nil {
traceStage(ctx, "read", err)
log.Warn(ctx, "Artwork: Failed to read resolved image", "kind", item.ItemKind, "id", item.ItemID, "source", res.source, err)
return outcomeFailed, nil, retryIn
}
log.Debug(ctx, "Artwork: Read resolved image", "kind", item.ItemKind, "id", item.ItemID,
"source", res.source, "bytes", len(data), "elapsed", time.Since(readStart))
hashStart := time.Now()
hash, err := hashImage(bytes.NewReader(data))
if err != nil {
traceStage(ctx, "hash", err)
log.Warn(ctx, "Artwork: Failed to hash image", "kind", item.ItemKind, "id", item.ItemID, err)
return outcomeFailed, nil, retryIn
}
log.Trace(ctx, "Artwork: Hashed image", "kind", item.ItemKind, "id", item.ItemID,
"hash", hash, "bytes", len(data), "elapsed", time.Since(hashStart))
art, err := repo.GetImage(hash)
switch {
case err == nil && art.Width > 0:
log.Debug(ctx, "Artwork: Reusing a known image, skipping decode", "kind", item.ItemKind,
"id", item.ItemID, "hash", hash)
// A row with no dimensions was stored when no decoder matched; retry in case one exists now.
case err == nil, errors.Is(err, model.ErrNotFound):
decodeStart := time.Now()
art, err = decodeArtwork(ctx, hash, data)
// Extension-matched local bytes we cannot decode are most likely a codec we lack; an
// external body carries no such guarantee, and empty bytes are no image at all.
if errors.Is(err, image.ErrFormat) && len(data) > 0 && isLocalSource(res.source) {
log.Debug(ctx, "Artwork: No decoder for this image format, storing it without placeholders",
"kind", item.ItemKind, "id", item.ItemID, "source", res.source, "bytes", len(data))
art, err = undecodedArtwork(hash), nil
}
if err != nil {
traceStage(ctx, "decode", err)
log.Warn(ctx, "Artwork: Failed to decode resolved image", "kind", item.ItemKind, "id", item.ItemID, err)
return outcomeFailed, nil, retryIn
}
log.Debug(ctx, "Artwork: Decoded new image", "kind", item.ItemKind, "id", item.ItemID, "hash", hash,
"width", art.Width, "height", art.Height, "mime", art.Mime, "elapsed", time.Since(decodeStart))
default:
traceStage(ctx, "lookup", err)
log.Warn(ctx, "Artwork: Failed to look up image hash", "kind", item.ItemKind, "id", item.ItemID, err)
return outcomeFailed, nil, retryIn
}
art.SizeBytes = int64(len(data))
ia, err := p.persist(ctx, repo, item, art, res, data)
if err != nil {
traceStage(ctx, "store", err)
log.Warn(ctx, "Artwork: Failed to persist resolved image", "kind", item.ItemKind, "id", item.ItemID, err)
return outcomeFailed, nil, retryIn
}
got = &acquired{ia: ia, mime: art.Mime, data: data}
if res.extErr != nil {
log.Debug(ctx, "Artwork: Serving a lower-priority source after an external failure",
"kind", item.ItemKind, "id", item.ItemID, "source", res.source)
return outcomeFoundStale, got, retryIn
}
return outcomeFound, got, retryIn
}
// persist places the bytes and commits the rows referencing them, excluding Prune for that
// window only so a slow resolution can never hold it off.
func (p *processor) persist(ctx context.Context, repo model.ArtworkRepository, item model.ArtworkQueueItem,
art *model.Artwork, res resolution, data []byte,
) (*model.ItemArtwork, error) {
if p.pruneLock != nil {
p.pruneLock.Lock()
defer p.pruneLock.Unlock()
}
sourcePath, refMtime, err := placeBytes(p.store, art, res, data)
if err != nil {
return nil, fmt.Errorf("writing image store: %w", err)
}
if err := repo.PutImage(art); err != nil {
return nil, fmt.Errorf("persisting artwork image: %w", err)
}
ia := &model.ItemArtwork{
ItemKind: item.ItemKind,
ItemID: item.ItemID,
ImageType: item.ImageType,
Hash: art.Hash,
Source: res.source,
SourcePath: sourcePath,
RefMtime: refMtime,
AttemptedAt: time.Now(),
Trace: traceFrom(ctx).encode(sourcePath),
}
// PutItemArtwork stamps UpdatedAt on ia, so the returned struct matches the persisted row.
if err := repo.PutItemArtwork(ia); err != nil {
return nil, fmt.Errorf("persisting item artwork state: %w", err)
}
return ia, nil
}
// writeAbsent records a known-absent state: every source answered definitively "no".
func writeAbsent(ctx context.Context, repo model.ArtworkRepository, item model.ArtworkQueueItem) outcome {
err := repo.PutItemArtwork(&model.ItemArtwork{
ItemKind: item.ItemKind,
ItemID: item.ItemID,
ImageType: item.ImageType,
AttemptedAt: time.Now(),
Trace: traceFrom(ctx).encode(""),
})
if err != nil {
log.Warn(ctx, "Artwork: Failed to persist absent state", "kind", item.ItemKind, "id", item.ItemID, err)
return outcomeFailed
}
log.Debug(ctx, "Artwork: Settled absent, every source answered definitively",
"kind", item.ItemKind, "id", item.ItemID)
return outcomeAbsent
}
func readCapped(r io.Reader) ([]byte, error) {
limit := maxImageBytes()
data, err := io.ReadAll(io.LimitReader(r, limit+1))
if err != nil {
return nil, err
}
if int64(len(data)) > limit {
return nil, fmt.Errorf("image exceeds size cap %d", limit)
}
return data, nil
}
// decodeCapped rejects declared dimensions over maxImagePixels before the full-decode allocation.
func decodeCapped(data []byte) (image.Image, string, error) {
cfg, format, err := image.DecodeConfig(bytes.NewReader(data))
if err != nil {
return nil, "", fmt.Errorf("decode image config: %w", err)
}
// Compared by division so the cap cannot be defeated by an int64 overflow.
if cfg.Width <= 0 || cfg.Height <= 0 || cfg.Width > maxImagePixels/cfg.Height {
return nil, "", fmt.Errorf("image dimensions %dx%d exceed pixel cap %d", cfg.Width, cfg.Height, maxImagePixels)
}
img, _, err := image.Decode(bytes.NewReader(data))
if err != nil {
return nil, "", fmt.Errorf("decode image: %w", err)
}
return img, format, nil
}
// undecodedArtwork is the row for bytes no decoder matched: servable, but with no dimensions
// and none of the placeholders a decode would have produced.
func undecodedArtwork(hash string) *model.Artwork {
return &model.Artwork{Hash: hash, Mime: mimeForFormat("")}
}
// decodeArtwork builds a new Artwork row from raw bytes: dimensions, mime and the two
// placeholder hashes, both encoded from one shared downscaled thumbnail.
func decodeArtwork(ctx context.Context, hash string, data []byte) (*model.Artwork, error) {
img, format, err := decodeCapped(data)
if err != nil {
return nil, err
}
thumb := makeThumbnail(img, thumbnailSize)
bh, err := blurhash.Encode(thumb)
if err != nil {
log.Warn(ctx, "Artwork: Blurhash encoding failed", "hash", hash, err)
bh = ""
}
var th string
if raw, err := thumbhash.Encode(thumb); err != nil {
log.Warn(ctx, "Artwork: Thumbhash encoding failed", "hash", hash, err)
} else {
th = base64.StdEncoding.EncodeToString(raw)
}
return &model.Artwork{
Hash: hash,
Mime: mimeForFormat(format),
Width: img.Bounds().Dx(),
Height: img.Bounds().Dy(),
BlurHash: bh,
ThumbHash: th,
DominantColor: dominant.Color(thumb),
}, nil
}
// makeThumbnail downscales img to fit within maxSize on its longest side; it never upscales.
func makeThumbnail(img image.Image, maxSize int) image.Image {
b := img.Bounds()
w, h := b.Dx(), b.Dy()
if w <= maxSize && h <= maxSize {
return toFastScaleType(img)
}
scale := float64(maxSize) / float64(max(w, h))
// NRGBA, not RGBA: thumbhash requires straight alpha, and blurhash reads this type without
// converting, so neither encoder allocates a second copy of the thumbnail.
dst := image.NewNRGBA(image.Rect(0, 0, max(1, int(float64(w)*scale)), max(1, int(float64(h)*scale))))
xdraw.CatmullRom.Scale(dst, dst.Bounds(), toFastScaleType(img), b, draw.Src, nil)
return dst
}
// isFileBacked reports whether the bytes already live in a library/upload file, so the
// content-addressed store must not duplicate them.
func isFileBacked(source string) bool {
return source == "folder" || source == "upload"
}
// isLocalSource reports whether the bytes came off disk rather than off the network.
func isLocalSource(source string) bool {
return isFileBacked(source) || source == "embedded"
}
// placeBytes reports the item's backing-file provenance and writes the bytes into the store
// for the sources that have none.
func placeBytes(store *ImageStore, art *model.Artwork, res resolution, data []byte) (sourcePath string, refMtime int64, err error) {
if isFileBacked(res.source) {
return res.sourcePath, res.refMtime, nil
}
if res.source == "embedded" {
sourcePath, refMtime = res.sourcePath, res.refMtime
}
return sourcePath, refMtime, store.Write(art.Hash, art.Mime, bytes.NewReader(data))
}
// mimeForFormat maps an image.Decode format name to its MIME type; extForMime is the inverse.
func mimeForFormat(format string) string {
switch format {
case "jpeg":
return "image/jpeg"
case "png":
return "image/png"
case "gif":
return "image/gif"
case "webp":
return "image/webp"
}
return "application/octet-stream"
}