Security:
- New internal/netguard package: podcast feed/enclosure fetching now
refuses to dial non-public IPs (loopback/RFC1918/link-local/cloud
metadata/CGNAT), validated per-connection including across redirects,
closing an SSRF gap where a malicious or compromised feed URL could
probe the server's internal network. Configurable via
VIBESONIC_PODCASTS_ALLOW_PRIVATE_NETWORKS for intentional LAN feeds.
- Podcast episode downloads capped at 4GB (was unbounded, disk-exhaustion risk)
- Cover art decoding now rejects images above a 40-megapixel budget
before full decode (decompression-bomb guard)
- HTTP server sets ReadHeaderTimeout (slowloris hardening)
- Admin folder-scan redirect now validated as same-site instead of
trusting the raw Referer header (open-redirect fix)
- Legacy Subsonic token/salt auth now uses a constant-time comparison
Web UI completion:
- Persistent player: shuffle and repeat (off/all/one) controls, wired
into queue advance/prev logic and persisted like other player state
- Admin user-permissions form: adminRole toggle (with a guard against
demoting the last remaining admin) and per-user music-folder access
checkboxes, backed by existing schema/enforcement that was previously
unreachable from the UI
- Playlist M3U8 import/export
Docs: added docs/client-compatibility.md (honest tracking template —
no client testing has been performed), docs/deployment.md,
docs/security.md, docs/backup.md; refreshed README and cross-links;
added a footnote in docs/subsonic-compatibility.md distinguishing
spec-standard from VibeSonic-original OpenSubsonic extensions.
Also added Go fuzz tests for the external-auth CIDR/IP parser and the
LRC lyrics parser.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014r98AtcBbevpscWLsfwVvv
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
owenandClaude Sonnet 5