navidrome/scanner/controller.go
Deluan Quintão b293b96256
refactor(persistence): stateless repositories with per-call context (#6149)
* refactor(persistence): adopt generic deluan/rest repository API

Pin deluan/rest to the refactor branch. REST-facing repository methods
take a context and return typed values. Drop DataStore.Resource and
ResourceRepository; the native API names typed repositories directly
through a per-request adapter that later commits remove.

* refactor(persistence): base repository helpers take a context

* refactor(persistence): LibraryRepository takes a context per call

* refactor(persistence): PropertyRepository takes a context per call

* refactor(persistence): UserPropsRepository takes a context per call

* refactor(persistence): TranscodingRepository takes a context per call

* refactor(persistence): ShareRepository takes a context per call

* refactor(persistence): PlayerRepository takes a context per call

* refactor(persistence): RadioRepository takes a context per call

* refactor(persistence): PlayQueueRepository takes a context per call

* refactor(persistence): Tag and Genre repositories take a context per call

* refactor(persistence): PluginRepository takes a context per call

* refactor(persistence): Scrobble repositories take a context per call

* refactor(persistence): FolderRepository takes a context per call

* refactor(persistence): Artwork repositories take a context per call

* refactor(persistence): UserRepository takes a context per call

* refactor(persistence): ArtistRepository takes a context per call

ReadAll no longer rewrites the shared sort mappings for the role filter;
it works on a per-call copy.

* test(persistence): assert artist role sort sanitization in ReadAll

* refactor(persistence): AlbumRepository takes a context per call

* test(persistence): pass the test context to album repository helpers

* refactor(persistence): MediaFileRepository takes a context per call

* refactor(persistence): Playlist repositories take a context per call

* refactor(persistence): build all repositories once per store

* refactor(core): REST repository wrappers are built once

* refactor(persistence): repositories are stateless

Remove the context field from the base repository and the per-request
REST adapter. Enable the containedctx linter so no repository can hold a
request context again.

* chore(lint): skip containedctx in test files

* refactor: share simplifications from the stateless repositories sweep

Add deleteOwnedAll on sqlRepository and use it in player/share Delete
to remove the duplicated bulk-delete loop; have Share.Repository()
return model.ShareRepository so subsonic sharing.go drops its repeated
type assertions.

* chore(core): assert REST wrappers implement Persistable

* chore: reformat imports

* perf(persistence): build repositories on first use

Each transaction store used to construct all 21 repositories up front,
paying for filter and sort mapping setup the block never touched. Fields
are now sync.OnceValue thunks, so a store only builds what it uses.

* fix(persistence): clean plugin references per deleted user

A bulk user delete that fails on a later id had already removed the
earlier rows but skipped their plugin cleanup. Cleanup now runs right
after each successful delete.

* fix(core): unload disabled plugins even when a user delete fails

A bulk delete can fail on a later id after earlier users were removed
and their plugins auto-disabled. The wrapper returned before unloading,
leaving those plugins running until the next successful delete or a
restart.

* chore(deps): pin deluan/rest to v1.0.1

Replaces the pseudo-version of the refactor branch with the tagged
release. REST error messages now name the bare type (Artist, not
model.Artist).

* test: use the spec context instead of context.Background()

Replace the context.Background()/context.TODO() calls this branch added
to tests with the spec's ctx, GinkgoT().Context(), or t/b.Context(), so
repository calls are bound to the running spec's lifetime.

* test: declare the spec context once per Describe

Set ctx from GinkgoT().Context() first in each top-level BeforeEach and reuse it, building user contexts on top of it instead of repeating inline calls.
2026-09-25 18:06:10 -04:00

399 lines
12 KiB
Go

package scanner
import (
"context"
"errors"
"fmt"
"slices"
"sync"
"sync/atomic"
"time"
"github.com/navidrome/navidrome/conf"
"github.com/navidrome/navidrome/consts"
"github.com/navidrome/navidrome/core/auth"
"github.com/navidrome/navidrome/core/metrics"
"github.com/navidrome/navidrome/core/playlists"
"github.com/navidrome/navidrome/db"
"github.com/navidrome/navidrome/log"
"github.com/navidrome/navidrome/model"
"github.com/navidrome/navidrome/model/request"
"github.com/navidrome/navidrome/server/events"
"github.com/navidrome/navidrome/utils/pl"
"github.com/navidrome/navidrome/utils/singleton"
"golang.org/x/time/rate"
)
var (
ErrAlreadyScanning = errors.New("already scanning")
)
func New(rootCtx context.Context, ds model.DataStore, broker events.Broker,
pls playlists.Playlists, m metrics.Metrics) model.Scanner {
c := &controller{
rootCtx: rootCtx,
ds: ds,
broker: broker,
pls: pls,
metrics: m,
devExternalScanner: conf.Server.DevExternalScanner,
}
if !c.devExternalScanner {
c.limiter = new(rate.Sometimes{Interval: conf.Server.DevActivityPanelUpdateRate})
}
return c
}
func (s *controller) getScanner() scanner {
if s.devExternalScanner {
return &scannerExternal{}
}
return &scannerImpl{ds: s.ds, pls: s.pls}
}
// CallScan starts an in-process scan of specific library/folder pairs.
// If targets is empty, it scans all libraries.
// This is meant to be called from the command line (see cmd/scan.go).
func CallScan(ctx context.Context, ds model.DataStore, pls playlists.Playlists, fullScan bool, targets []model.ScanTarget) (<-chan *ProgressInfo, error) {
release, err := lockScan(ctx)
if err != nil {
return nil, err
}
defer release()
ctx = auth.WithAdminUser(ctx, ds)
progress := make(chan *ProgressInfo, 100)
go func() {
defer close(progress)
scanner := &scannerImpl{ds: ds, pls: pls}
scanner.scanFolders(ctx, fullScan, targets, progress)
}()
return progress, nil
}
func IsScanning() bool {
return running.Load()
}
type ProgressInfo struct {
LibID int
FileCount uint32
Path string
Phase string
ChangesDetected bool
Warning string
Error string
ForceUpdate bool
}
// scanner defines the interface for different scanner implementations.
// This allows for swapping between in-process and external scanners.
type scanner interface {
// scanFolders performs the actual scanning of folders. If targets is nil, it scans all libraries.
scanFolders(ctx context.Context, fullScan bool, targets []model.ScanTarget, progress chan<- *ProgressInfo)
}
type controller struct {
rootCtx context.Context //nolint:containedctx // scanner lifecycle ctx
ds model.DataStore
broker events.Broker
metrics metrics.Metrics
pls playlists.Playlists
limiter *rate.Sometimes
devExternalScanner bool
count atomic.Uint32
folderCount atomic.Uint32
changesDetected bool
}
// getLastScanTime returns the most recent scan time across all libraries
func (s *controller) getLastScanTime(ctx context.Context) (time.Time, error) {
libs, err := s.ds.Library().GetAll(ctx, model.QueryOptions{
Sort: "last_scan_at",
Order: "desc",
Max: 1,
})
if err != nil {
return time.Time{}, fmt.Errorf("getting libraries: %w", err)
}
if len(libs) == 0 {
return time.Time{}, nil
}
return libs[0].LastScanAt, nil
}
// getScanInfo retrieves scan status from the database
func (s *controller) getScanInfo(ctx context.Context) (scanType string, elapsed time.Duration, lastErr string) {
lastErr, _ = s.ds.Property().DefaultGet(ctx, consts.LastScanErrorKey, "")
scanType, _ = s.ds.Property().DefaultGet(ctx, consts.LastScanTypeKey, "")
startTimeStr, _ := s.ds.Property().DefaultGet(ctx, consts.LastScanStartTimeKey, "")
if startTimeStr != "" {
startTime, err := time.Parse(time.RFC3339, startTimeStr)
if err == nil {
if running.Load() {
elapsed = time.Since(startTime)
} else {
// If scan is not running, calculate elapsed time using the most recent scan time
lastScanTime, err := s.getLastScanTime(ctx)
if err == nil && !lastScanTime.IsZero() {
elapsed = lastScanTime.Sub(startTime)
}
}
}
}
return scanType, elapsed, lastErr
}
func (s *controller) Status(ctx context.Context) (*model.ScannerStatus, error) {
lastScanTime, err := s.getLastScanTime(ctx)
if err != nil {
return nil, fmt.Errorf("getting last scan time: %w", err)
}
scanType, elapsed, lastErr := s.getScanInfo(ctx)
if running.Load() {
status := &model.ScannerStatus{
Scanning: true,
LastScan: lastScanTime,
Count: s.count.Load(),
FolderCount: s.folderCount.Load(),
LastError: lastErr,
ScanType: scanType,
ElapsedTime: elapsed,
}
return status, nil
}
count, folderCount, err := s.getCounters(ctx)
if err != nil {
return nil, fmt.Errorf("getting library stats: %w", err)
}
return &model.ScannerStatus{
Scanning: false,
LastScan: lastScanTime,
Count: uint32(count),
FolderCount: uint32(folderCount),
LastError: lastErr,
ScanType: scanType,
ElapsedTime: elapsed,
}, nil
}
func (s *controller) getCounters(ctx context.Context) (int64, int64, error) {
libs, err := s.ds.Library().GetAll(ctx)
if err != nil {
return 0, 0, fmt.Errorf("library count: %w", err)
}
var count, folderCount int64
for _, l := range libs {
count += int64(l.TotalSongs)
folderCount += int64(l.TotalFolders)
}
return count, folderCount, nil
}
func (s *controller) ScanAll(requestCtx context.Context, fullScan bool) ([]string, error) {
return s.ScanFolders(requestCtx, fullScan, nil)
}
func (s *controller) ScanFolders(requestCtx context.Context, fullScan bool, targets []model.ScanTarget) ([]string, error) {
release, err := lockScan(requestCtx)
if err != nil {
return nil, err
}
defer release()
// Prepare the context for the scan
ctx := request.AddValues(s.rootCtx, requestCtx)
ctx = auth.WithAdminUser(ctx, s.ds)
// A quick scan is promoted to a full one when it resumes an interrupted full scan; that happens
// inside the scanner (possibly in a subprocess), so mirror it here for the analysis gate. Must
// be read before the scan: ScanEnd clears the flag.
effectiveFullScan := EffectiveFullScan(ctx, s.ds, fullScan, targets)
if effectiveFullScan || s.includesUnscannedLibrary(ctx, targets) {
if err := db.MarkOptimizePending(ctx); err != nil {
log.Error(ctx, "Scanner: Error marking DB analysis pending", err)
}
}
// Send the initial scan status event
s.sendMessage(ctx, &events.ScanStatus{Scanning: true, Count: 0, FolderCount: 0})
progress := make(chan *ProgressInfo, 100)
go func() {
defer close(progress)
scanner := s.getScanner()
scanner.scanFolders(ctx, fullScan, targets, progress)
}()
// Wait for the scan to finish, sending progress events to all connected clients
scanWarnings, scanError := s.trackProgress(ctx, progress)
for _, w := range scanWarnings {
log.Warn(ctx, fmt.Sprintf("Scan warning: %s", w))
}
// Store scan error in database so it can be displayed in the UI
if scanError != nil {
_ = s.ds.Property().Put(ctx, consts.LastScanErrorKey, scanError.Error())
}
// Refresh the query-planner statistics after a successful full scan. This must run in the
// server process: with the external scanner, an ANALYZE in the subprocess is invisible to the
// server's pooled connections; their shared schema cache keeps the old statistics until the
// process restarts.
if effectiveFullScan && scanError == nil {
if err := db.Optimize(ctx); err != nil {
log.Error(ctx, "Scanner: Error analyzing DB", err)
}
}
// If changes were detected, send a refresh event to all clients
if s.changesDetected {
log.Debug(ctx, "Library changes imported. Sending refresh event")
s.broker.SendBroadcastMessage(ctx, &events.RefreshResource{})
}
// Send the final scan status event, with totals
if count, folderCount, err := s.getCounters(ctx); err != nil {
s.metrics.WriteAfterScanMetrics(ctx, false)
return scanWarnings, err
} else {
scanType, elapsed, lastErr := s.getScanInfo(ctx)
s.metrics.WriteAfterScanMetrics(ctx, true)
s.sendMessage(ctx, &events.ScanStatus{
Scanning: false,
Count: count,
FolderCount: folderCount,
Error: lastErr,
ScanType: scanType,
ElapsedTime: elapsed,
})
}
return scanWarnings, scanError
}
// This is a global variable that is used to prevent multiple scans from running at the same time.
// "There can be only one" - https://youtu.be/sqcLjcSloXs?si=VlsjEOjTJZ68zIyg
var (
running atomic.Bool
scanMaintenanceMux sync.Mutex
)
func lockScan(ctx context.Context) (func(), error) {
if !running.CompareAndSwap(false, true) {
log.Debug(ctx, "Scanner already running, ignoring request")
return func() {}, ErrAlreadyScanning
}
scanMaintenanceMux.Lock()
return func() {
scanMaintenanceMux.Unlock()
running.Store(false)
}, nil
}
// LockForMaintenance prevents a scan from starting while database maintenance is running.
func LockForMaintenance() (func(), bool) {
if !scanMaintenanceMux.TryLock() {
return func() {}, false
}
if running.Load() {
scanMaintenanceMux.Unlock()
return func() {}, false
}
return scanMaintenanceMux.Unlock, true
}
// EffectiveFullScan reports whether a scan was requested as full or will resume an interrupted
// full scan in one of the included libraries.
func EffectiveFullScan(ctx context.Context, ds model.DataStore, fullScan bool, targets []model.ScanTarget) bool {
if fullScan {
return true
}
return anyIncludedLibrary(ctx, ds, targets, func(library model.Library) bool {
return library.FullScanInProgress
})
}
func (s *controller) includesUnscannedLibrary(ctx context.Context, targets []model.ScanTarget) bool {
return anyIncludedLibrary(ctx, s.ds, targets, func(library model.Library) bool {
return library.LastScanAt.IsZero()
})
}
// anyIncludedLibrary reports whether any library included in the scan (all of them when targets is
// empty) matches pred.
func anyIncludedLibrary(ctx context.Context, ds model.DataStore, targets []model.ScanTarget, pred func(model.Library) bool) bool {
libraries, err := ds.Library().GetAll(ctx)
if err != nil {
return false
}
if len(targets) == 0 {
return slices.ContainsFunc(libraries, pred)
}
targeted := make(map[int]struct{}, len(targets))
for _, target := range targets {
targeted[target.LibraryID] = struct{}{}
}
return slices.ContainsFunc(libraries, func(library model.Library) bool {
_, ok := targeted[library.ID]
return ok && pred(library)
})
}
func (s *controller) trackProgress(ctx context.Context, progress <-chan *ProgressInfo) ([]string, error) {
s.count.Store(0)
s.folderCount.Store(0)
s.changesDetected = false
var warnings []string
var errs []error
for p := range pl.ReadOrDone(ctx, progress) {
if p.Error != "" {
errs = append(errs, errors.New(p.Error))
continue
}
if p.Warning != "" {
warnings = append(warnings, p.Warning)
continue
}
if p.ChangesDetected {
s.changesDetected = true
continue
}
s.count.Add(p.FileCount)
if p.FileCount > 0 {
s.folderCount.Add(1)
}
scanType, elapsed, lastErr := s.getScanInfo(ctx)
status := &events.ScanStatus{
Scanning: true,
Count: int64(s.count.Load()),
FolderCount: int64(s.folderCount.Load()),
Error: lastErr,
ScanType: scanType,
ElapsedTime: elapsed,
}
if s.limiter != nil && !p.ForceUpdate {
s.limiter.Do(func() { s.sendMessage(ctx, status) })
} else {
s.sendMessage(ctx, status)
}
}
return warnings, errors.Join(errs...)
}
func (s *controller) sendMessage(ctx context.Context, status *events.ScanStatus) {
s.broker.SendBroadcastMessage(ctx, status)
}
// GetInstance returns the scanner singleton: Status reads the progress counters of the controller
// running the scan, and scheduler, watcher and signal scans do not start from the API's injector.
func GetInstance(rootCtx context.Context, ds model.DataStore, broker events.Broker,
pls playlists.Playlists, m metrics.Metrics) model.Scanner {
return singleton.GetInstance(func() *controller {
return New(rootCtx, ds, broker, pls, m).(*controller)
})
}