Files
owenandClaude Sonnet 5 b6acd63c2d Implement sonicSimilarity extension with real audio feature analysis
Adds internal/audiofeatures: decodes each scanned song to PCM via
ffmpeg and computes actual measured characteristics — tempo (via
autocorrelation of the frame energy envelope, left unset rather than
guessed when a clip is too short to contain a beat), an approximate
loudness figure, spectral centroid/rolloff and zero-crossing rate (via
a hand-written FFT), and a 12-bin chroma vector used both for a
Krumhansl-Kessler key estimate and as similarity-index input. No
external DSP/ML dependency.

The scanner now runs this analysis after indexing each song (as part
of the already-background scan, not on any request path) and stores
results in the existing audio_features table via the new
database.UpsertAudioFeatures/GetAudioFeatures/ListAllAudioFeatures.

internal/similarity is a small in-process nearest-neighbor index:
z-score normalizes each feature dimension across the analyzed library
and answers Nearest (ranked real distances) and Path (a feature-space
straight-line walk from one track to another, hopping to the closest
not-yet-used real song at each step — never a fabricated intermediate
track).

getSonicSimilarTracks and findSonicPath expose these through the
standard sonicMatch response, converting distance to a normalized
(0,1] similarity score. A song that hasn't been analyzed yet is simply
absent from results rather than faked.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014r98AtcBbevpscWLsfwVvv
2026-08-23 08:30:20 -05:00

337 lines
10 KiB
Go

// Package scanner walks configured music folders, detects new/changed/
// deleted/renamed files, and indexes them into the library via a bounded
// worker pool.
package scanner
import (
"context"
"database/sql"
"fmt"
"io/fs"
"log/slog"
"os"
"path/filepath"
"strings"
"sync"
"time"
"git.noroutes.net/NoRoutes/vibetunes/internal/audiofeatures"
"git.noroutes.net/NoRoutes/vibetunes/internal/coverart"
"git.noroutes.net/NoRoutes/vibetunes/internal/database"
"git.noroutes.net/NoRoutes/vibetunes/internal/library"
"git.noroutes.net/NoRoutes/vibetunes/internal/metadata"
"git.noroutes.net/NoRoutes/vibetunes/internal/streaming"
)
// recognizedExtensions are the file extensions VibeSonic attempts to
// probe as media. Anything else (art, .nfo, playlists, ...) is skipped
// during a scan rather than passed to ffprobe.
var recognizedExtensions = map[string]bool{
".mp3": true, ".flac": true, ".ogg": true, ".oga": true, ".opus": true,
".aac": true, ".m4a": true, ".m4b": true, ".alac": true,
".wav": true, ".aiff": true, ".aif": true, ".wma": true, ".mp4": true,
}
const maxWorkers = 4
// Scanner indexes one or more music folders into the database.
type Scanner struct {
DB *sql.DB
Prober *metadata.Prober
Log *slog.Logger
FFmpegPath string // for sonic-analysis audio decoding; "" defaults to "ffmpeg" on PATH
mu sync.Mutex
scanning map[string]bool // music_folder_id -> in progress
}
func New(db *sql.DB, prober *metadata.Prober, log *slog.Logger) *Scanner {
return &Scanner{DB: db, Prober: prober, Log: log, scanning: make(map[string]bool)}
}
// ScanFolder walks a single music folder and indexes it incrementally:
// unchanged files (same size+mtime) are only marked seen, changed files
// are reprobed, moved files are matched by content hash and just get
// their path updated, and files no longer found on disk are marked
// missing rather than deleted outright.
func (s *Scanner) ScanFolder(ctx context.Context, folderID string) error {
if !s.beginScan(folderID) {
return fmt.Errorf("scan already in progress for folder %s", folderID)
}
defer s.endScan(folderID)
folder, err := database.GetMusicFolder(s.DB, folderID)
if err != nil {
return err
}
startedAt := time.Now()
if err := database.StartScan(s.DB, folderID); err != nil {
return err
}
var seen, added, updated int
var mu sync.Mutex
sem := make(chan struct{}, maxWorkers)
var wg sync.WaitGroup
var firstErr error
walkErr := filepath.WalkDir(folder.Path, func(path string, d fs.DirEntry, err error) error {
if err != nil {
s.Log.Warn("scan: walk error", "path", path, "err", err)
return nil // keep scanning the rest of the tree
}
if ctx.Err() != nil {
return ctx.Err()
}
if d.IsDir() {
return nil
}
if !recognizedExtensions[strings.ToLower(filepath.Ext(path))] {
return nil
}
rel, err := filepath.Rel(folder.Path, path)
if err != nil {
return nil
}
rel = filepath.ToSlash(rel)
wg.Add(1)
sem <- struct{}{}
go func() {
defer wg.Done()
defer func() { <-sem }()
wasAdded, wasUpdated, err := s.indexOne(ctx, folder.ID, path, rel)
mu.Lock()
defer mu.Unlock()
seen++
if err != nil {
s.Log.Warn("scan: index failed", "path", rel, "err", err)
} else {
if wasAdded {
added++
} else if wasUpdated {
updated++
}
}
if seen%25 == 0 {
_ = database.UpdateScanProgress(s.DB, folder.ID, seen, added, updated)
}
}()
return nil
})
wg.Wait()
if walkErr != nil && firstErr == nil {
firstErr = walkErr
}
_ = database.UpdateScanProgress(s.DB, folder.ID, seen, added, updated)
removed, err := database.MarkFolderFilesMissingExcept(s.DB, folder.ID, startedAt)
if err != nil && firstErr == nil {
firstErr = err
}
errMsg := ""
if firstErr != nil {
errMsg = firstErr.Error()
}
if err := database.FinishScan(s.DB, folder.ID, int(removed), errMsg); err != nil {
return err
}
return firstErr
}
// indexOne probes and indexes a single file, returning whether it was
// newly added or updated so the caller can track progress.
func (s *Scanner) indexOne(ctx context.Context, folderID, absPath, relPath string) (added, updated bool, err error) {
info, err := os.Stat(absPath)
if err != nil {
return false, false, err
}
size := info.Size()
mtime := info.ModTime().Unix()
hash := fingerprint(size, mtime)
existing, err := database.FindMediaFileByPath(s.DB, folderID, relPath)
if err == nil {
if existing.SizeBytes == size && existing.MtimeUnix == mtime {
return false, false, database.TouchMediaFileScanned(s.DB, existing.ID)
}
// Changed: reprobe and update in place; the media_file (and
// therefore song) ID stays the same.
tags, perr := s.Prober.Probe(ctx, absPath)
if perr != nil {
return false, false, perr
}
if err := database.UpdateMediaFileScanned(s.DB, existing.ID, size, mtime, hash, tags.Format, tags.Kind, tags.DurationSeconds, tags.BitrateKbps, tags.SampleRateHz, tags.Channels); err != nil {
return false, false, err
}
mf, err := database.GetMediaFile(s.DB, existing.ID)
if err != nil {
return false, false, err
}
song, err := library.IndexFile(s.DB, mf, absPath, tags)
if err != nil {
return false, false, err
}
s.indexCoverArt(absPath, song, tags)
s.analyzeAudio(ctx, absPath, song, tags)
return false, true, nil
}
if err != database.ErrNotFound {
return false, false, err
}
// Not found by path. Check whether this is a rename/move of a file we
// already indexed (same content fingerprint, different path).
if renamed, rerr := database.FindMediaFileByHash(s.DB, folderID, hash); rerr == nil {
if err := database.RenameMediaFile(s.DB, renamed.ID, relPath); err != nil {
return false, false, err
}
return false, true, nil
}
// Genuinely new file.
tags, err := s.Prober.Probe(ctx, absPath)
if err != nil {
return false, false, err
}
mf, err := database.CreateMediaFile(s.DB, folderID, relPath, size, mtime, hash, tags.Format, tags.Kind, tags.DurationSeconds, tags.BitrateKbps, tags.SampleRateHz, tags.Channels)
if err != nil {
return false, false, err
}
song, err := library.IndexFile(s.DB, mf, absPath, tags)
if err != nil {
return false, false, err
}
s.indexCoverArt(absPath, song, tags)
s.analyzeAudio(ctx, absPath, song, tags)
return true, false, nil
}
// analyzeAudio runs the sonic-feature analyzer over a scanned file and
// stores the result, powering the sonicSimilarity extension. This runs
// as part of the (already background, admin-triggered) scan itself
// rather than on any request path. A decode/analysis failure is logged
// and otherwise ignored: the song is still indexed and playable, it
// just won't appear in sonic-similarity results until a future rescan
// succeeds — never a reason to fail the whole scan.
func (s *Scanner) analyzeAudio(ctx context.Context, absPath string, song *database.Song, tags *metadata.Tags) {
if tags.Kind != "song" {
return
}
features, err := audiofeatures.Analyze(ctx, s.FFmpegPath, absPath)
if err != nil {
s.Log.Warn("audio feature analysis failed", "err", err, "path", absPath)
return
}
if err := database.UpsertAudioFeatures(s.DB, song.ID, features); err != nil {
s.Log.Warn("storing audio features failed", "err", err, "path", absPath)
}
}
// indexCoverArt records where an album's cover art can be found: a
// folder image next to the file if one exists, otherwise the file's own
// embedded artwork (extracted lazily by the web layer on request, not
// here). Failures are non-fatal to the scan.
func (s *Scanner) indexCoverArt(absPath string, song *database.Song, tags *metadata.Tags) {
if !song.AlbumID.Valid {
return
}
dir := filepath.Dir(absPath)
if folderImage := coverart.FindFolderImage(dir); folderImage != "" {
hash, err := coverart.ContentHashOf(folderImage)
if err != nil {
return
}
_, _ = database.UpsertAlbumCoverArt(s.DB, song.AlbumID.String, "folder", folderImage, "", streaming.MimeType(folderImage), hash, 0, 0)
return
}
if tags.HasEmbeddedArt {
_, _ = database.UpsertAlbumCoverArt(s.DB, song.AlbumID.String, "embedded", "", song.MediaFileID, "image/jpeg", song.MediaFileID, 0, 0)
}
}
// fingerprint is a cheap, fast rename/change detector based on size and
// mtime rather than hashing file content (which would mean reading every
// byte of every file on every scan).
func fingerprint(size, mtime int64) string {
return fmt.Sprintf("%d-%d", size, mtime)
}
func (s *Scanner) beginScan(folderID string) bool {
s.mu.Lock()
defer s.mu.Unlock()
if s.scanning[folderID] {
return false
}
s.scanning[folderID] = true
return true
}
func (s *Scanner) endScan(folderID string) {
s.mu.Lock()
defer s.mu.Unlock()
delete(s.scanning, folderID)
}
// IsScanning reports whether a scan of folderID is currently running.
func (s *Scanner) IsScanning(folderID string) bool {
s.mu.Lock()
defer s.mu.Unlock()
return s.scanning[folderID]
}
// IsAnyScanning reports whether any folder is currently being scanned.
func (s *Scanner) IsAnyScanning() bool {
s.mu.Lock()
defer s.mu.Unlock()
return len(s.scanning) > 0
}
// ScanAllEnabled scans every enabled music folder sequentially (folders
// run one at a time; files within a folder run concurrently via the
// worker pool above).
func (s *Scanner) ScanAllEnabled(ctx context.Context) error {
folders, err := database.ListMusicFolders(s.DB)
if err != nil {
return err
}
var firstErr error
for _, f := range folders {
if !f.Enabled {
continue
}
if err := s.ScanFolder(ctx, f.ID); err != nil && firstErr == nil {
firstErr = err
}
}
return firstErr
}
// RunScheduled runs ScanAllEnabled once at startup and then on the given
// interval, until ctx is cancelled. interval <= 0 disables the periodic
// rescan (startup scan still runs).
func (s *Scanner) RunScheduled(ctx context.Context, interval time.Duration) {
if err := s.ScanAllEnabled(ctx); err != nil {
s.Log.Warn("startup scan finished with errors", "err", err)
}
if interval <= 0 {
return
}
ticker := time.NewTicker(interval)
defer ticker.Stop()
for {
select {
case <-ctx.Done():
return
case <-ticker.C:
if err := s.ScanAllEnabled(ctx); err != nil {
s.Log.Warn("scheduled scan finished with errors", "err", err)
}
}
}
}