refactor(db): migrate all ids to a uniform canonical 128-bit base62 encoding (#5824)

* refactor(model): extract canonical 128-bit base62 id codec

* feat(model): generate random ids as canonical 128-bit base62 values

* feat(scanner): emit legacy PIDs in canonical base62 encoding

* feat(db): add id canonicalization transform for the uniform-ids migration

* feat(db): migrate all ids to canonical 128-bit base62 encoding

* fix(db): canonicalize ids in junction tables and JSON columns

* chore(jellyfin): update id-family notes for uniform canonical ids

* test(ids): harden codec input contract and migration edge coverage

* refactor(model): use log.Fatal for Encode128 contract guard per project convention

* fix(db): force full rescan after id migration for legacy PID configs

* test(db): guard id-column inventory against schema drift

* refactor(ids): compile-time Encode128 contract and unified column rewrite helper

* refactor(db): apply review feedback to id migration

Filter empty strings in collectColumn's SQL, reuse a prepared statement
for rewriteColumn updates, and clarify the legacy ID functions' comment
now that they emit the canonical encoding.

* feat(auth): split session and public-link JWT secrets, rotating sessions on id migration

* test(subsonic): initialize public token secret in helpers suite

The suite sets auth.TokenAuth directly instead of calling auth.Init, so the
new PublicTokenAuth was nil whenever Ginkgo's spec order ran a helpers spec
before any spec that calls auth.Init, panicking in publicurl.ImageURL.

* refactor(db): inline canonicalID into its only consumer, the uniform-ids migration

* refactor(model): rename Encode128/Decode128 to Encode/Decode

With every id now exactly 128 bits, the width suffix is redundant; the
package-qualified id.Encode/id.Decode carries the same information.

* test(db): make the id-columns guard classify JSON columns too

The guard only inspected columns named id/pid/*_id, so it could not see ids
embedded in JSON. Widen it to *_ids and to every JSON column, and drive the
"covered" set from a new embeddedIDColumns list instead of the inline calls
in the migration.

Every JSON column the schema has now carries a verdict. The four denormalized
caches -- media_file/album.participants, media_file/album.tags,
album.folder_ids and artist.similar_artists -- hold only artist, tag and
folder ids. Those all come from id.NewHash, whose 22-char base62 encoding of
a 128-bit MD5 is already in canonical range, so canonicalID is the identity
on them and the migration correctly leaves them alone. A new codec test pins
that invariant, since the exemptions depend on it.

Verified on a copy of a 727MB/96k-track production database: canonicalizing
those four columns changed zero rows, and artist, tag and folder ids were
themselves unchanged by the migration (only media_file ids moved, 95108 of
96666).

server/jellyfin/README.md

@@ -90,9 +90,9 @@ skipped, so it doesn't create a nameless player.
 Navidrome item ids are **hex-encoded at the API boundary** (`dto.EncodeID`/`DecodeID`): every id
 is hex-encoded on the way out and hex-decoded on the way in. This is required because some clients
 parse ids as radix-16 — Finamp's queue `packIds`, for instance, does `int.parse(chunk, radix:16)`,
-which chokes on Navidrome's base-62 nanoids (e.g. `5QFKvMsJrd57QE2Le2dKKo`). Because a raw MD5 id
-from an old migrated library is itself valid hex, correctness depends on every emit path encoding
-and every receive path decoding — see `dto/ids.go`.
+which chokes on Navidrome's base62 ids (e.g. `5QFKvMsJrd57QE2Le2dKKo`). Because a base62 id can
+itself be valid hex, correctness depends on every emit path encoding and every receive path
+decoding — see `dto/ids.go`.
 
 ## Multi-library behavior
 
@@ -184,13 +184,13 @@ their Navidrome `ArtworkID`.
 
 Real Jellyfin item ids are GUIDs — 128-bit values, always 32 hex characters. Finamp relies on that
 when persisting its play queue across restarts: `packIds()` bit-packs every id into exactly 16
-bytes. Navidrome ids are longer (nanoid ids can exceed 128 bits, so they cannot be mapped into
-GUIDs), which means Finamp silently stores only the first 16 characters of each id and asks for
-those **truncated ids** back when restoring the queue — item lookups, then streaming, images,
-favorites and playback reports for the restored tracks.
+bytes. Navidrome ids are 22-character base62 strings, not 32-hex GUIDs, which means Finamp silently
+stores only the first 16 characters of each id and asks for those **truncated ids** back when
+restoring the queue — item lookups, then streaming, images, favorites and playback reports for the
+restored tracks.
 
 This API compensates server-side (`truncated_ids.go`): a 16-character id — a length no Navidrome
-id family uses — is resolved to the full id by unique-prefix lookup (an indexed range scan;
+id uses — is resolved to the full id by unique-prefix lookup (an indexed range scan;
 ambiguity is detected and fails safe). The `/Items?ids=` batch response echoes the id **as
 requested**, because Finamp matches restored items back to its stored ids, and the other item
 endpoints accept truncated ids transparently.

server/jellyfin/dto/ids.go

@@ -3,7 +3,7 @@ package dto
 import "encoding/hex"
 
 // EncodeID renders a Navidrome id as lowercase hex; Jellyfin clients parse ids as radix-16 (e.g.
-// Finamp's queue packing) and crash on Navidrome's base62 nanoids if emitted as-is.
+// Finamp's queue packing) and crash on Navidrome's base62 ids if emitted as-is.
 func EncodeID(id string) string {
 	if id == "" {
 		return ""

server/jellyfin/dto/ids_test.go

@@ -6,7 +6,7 @@ import (
 )
 
 var _ = Describe("id codec", func() {
-	It("round-trips a base62 nanoid through Encode/Decode", func() {
+	It("round-trips a base62 id", func() {
 		id := "5QFKvMsJrd57QE2Le2dKKo"
 		Expect(DecodeID(EncodeID(id))).To(Equal(id))
 	})

server/jellyfin/truncated_ids.go

@@ -11,8 +11,8 @@ import (
 )
 
 // truncatedIDLen is what Finamp's saved-queue persistence cuts item ids to (16 bytes, assuming
-// Jellyfin GUIDs). No Navidrome id family is 16 chars (nanoid=22, legacy MD5=32, playlist
-// UUID=36), so the length alone identifies a truncated id. See README.
+// Jellyfin GUIDs). All Navidrome ids are 22 chars (share ids 10), so length alone flags a
+// truncated id. See README.
 //
 // Handlers taking an item id resolve it via resolveItemID/resolveItemIDs; playlist-write handlers
 // and ParentId scoping don't (a restored queue never edits playlists or browses by container id).

server/public/handle_images.go

@@ -68,7 +68,7 @@ func (pub *Router) handleImages(w http.ResponseWriter, r *http.Request) {
 }
 
 func decodeArtworkID(tokenString string) (model.ArtworkID, error) {
-	token, err := auth.TokenAuth.Decode(tokenString)
+	token, err := auth.PublicTokenAuth.Decode(tokenString)
 	if err != nil {
 		return model.ArtworkID{}, err
 	}

server/public/handle_images_test.go

@@ -9,7 +9,7 @@ import (
 
 var _ = Describe("decodeArtworkID", func() {
 	BeforeEach(func() {
-		auth.TokenAuth = jwtauth.New("HS256", []byte("super secret"), nil)
+		auth.PublicTokenAuth = jwtauth.New("HS256", []byte("super secret"), nil)
 	})
 
 	It("fails to decode an invalid token", func() {

server/public/handle_shares.go

@@ -111,8 +111,8 @@ func (pub *Router) mapShareToM3U(r *http.Request, s model.Share) *model.Share {
 // admin flag) and grants access to nothing beyond the share it belongs to; the
 // stream handler still verifies the share exists, is unexpired, and that the
 // track is actually a member of it. An attacker who can forge these tokens
-// necessarily already holds the signing secret, which also signs real user
-// sessions, so that scenario is out of scope for the share boundary specifically.
+// necessarily already holds the public-link signing secret, a full-server
+// compromise that is out of scope for the share boundary specifically.
 func encodeMediafileShare(s model.Share, id string) string {
 	claims := auth.Claims{
 		ID:      id,

server/public/handle_streams.go

@@ -107,7 +107,7 @@ func shareContainsTrack(share *model.Share, mediaFileID string) bool {
 // public-share capability, not an auth credential; see encodeMediafileShare for
 // why a JWT is used here.
 func decodeStreamInfo(tokenString string) (shareTrackInfo, error) {
-	c, err := auth.Validate(tokenString)
+	c, err := auth.ValidatePublic(tokenString)
 	if err != nil {
 		return shareTrackInfo{}, err
 	}

server/public/handle_streams_test.go

@@ -29,7 +29,7 @@ func (m *mockStreamer) NewStream(_ context.Context, _ *model.MediaFile, r stream
 
 var _ = Describe("decodeStreamInfo", func() {
 	BeforeEach(func() {
-		auth.TokenAuth = jwtauth.New("HS256", []byte("test-secret"), nil)
+		auth.PublicTokenAuth = jwtauth.New("HS256", []byte("test-secret"), nil)
 	})
 
 	It("decodes a valid token with all fields", func() {
@@ -81,7 +81,7 @@ var _ = Describe("decodeStreamInfo", func() {
 
 var _ = Describe("encodeMediafileShare", func() {
 	BeforeEach(func() {
-		auth.TokenAuth = jwtauth.New("HS256", []byte("test-secret"), nil)
+		auth.PublicTokenAuth = jwtauth.New("HS256", []byte("test-secret"), nil)
 	})
 
 	It("includes the share ID in the token", func() {
@@ -113,7 +113,7 @@ var _ = Describe("handleStream", func() {
 	var pub *Router
 
 	BeforeEach(func() {
-		auth.TokenAuth = jwtauth.New("HS256", []byte("test-secret"), nil)
+		auth.PublicTokenAuth = jwtauth.New("HS256", []byte("test-secret"), nil)
 		ds = &tests.MockDataStore{}
 		shareRepo = &tests.MockShareRepo{}
 		ds.MockedShare = shareRepo

server/subsonic/helpers_test.go

@@ -22,6 +22,7 @@ var _ = Describe("helpers", func() {
 	BeforeEach(func() {
 		DeferCleanup(configtest.SetupConfig())
 		auth.TokenAuth = jwtauth.New("HS256", []byte("test secret"), nil)
+		auth.PublicTokenAuth = jwtauth.New("HS256", []byte("test public secret"), nil)
 	})
 
 	Describe("fakePath", func() {