Riverpod FutureProvider that reads the caller's system-playlists
build state. Consumed by the home Playlists row to decide between
real and placeholder cards.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds the model + MeApi.systemPlaylistsStatus() for the home Playlists
row's placeholder-variant logic (building / failed / pending /
seed-needed). Mirrors GET /api/me/system-playlists-status which
returns the caller's most recent system_playlist_runs row.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The server has always returned an enveloped response from GET
/api/playlists; the existing client decoded it as a flat List which
fails at runtime against the actual Map shape. Existing playlists
list screen would have been broken on any production instance.
list() now returns PlaylistsList { owned, public }. The existing
list screen consumes lists.all (owned + public flattened) — same
visible output as the previous flat-list assumption.
This unblocks the home parity slice which needs the same endpoint
for the new Playlists row.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Audio playback failed against the prod HTTPS server with
"Cleartext HTTP traffic to 127.0.0.1 not permitted". Diagnostic
logging confirmed _baseUrl was correctly set to the prod URL — the
127.0.0.1 wasn't coming from the server or our URL construction.
Root cause: just_audio + audio_service spin up a local HTTP proxy on
loopback to inject the Authorization header, because Android's
MediaSession API can't pass headers down to ExoPlayer directly.
ExoPlayer connects to http://127.0.0.1:<random-port>; the proxy adds
the Bearer header and forwards to the real upstream HTTPS URL.
Loopback traffic doesn't leave the device.
This is the documented just_audio happy path — see
https://pub.dev/packages/just_audio#a-note-on-android-cleartext-traffic
— not a workaround for a plugin bug.
Adds network_security_config.xml that scopes cleartext permission to
127.0.0.1 ONLY. The base config keeps cleartextTrafficPermitted=false
so any actual remote endpoint must still be HTTPS.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two issues from on-device testing against prod HTTPS:
1. ServerImage was resolving cover URLs correctly
(https://minstrel.fabledsword.com/api/albums/.../cover) but the server
returned 401 because Image.network doesn't carry the session token
automatically the way the browser sends cookies. Forwards the stored
session token as an Authorization: Bearer header. No-op when no token
is present.
2. The "Cleartext HTTP traffic to 127.0.0.1" audio error reproduced even
after the previous defensive check landed, which means the URL handed
to ExoPlayer has a valid scheme+host (just the wrong host). The check
only catches scheme-less URLs, so it didn't fire. Added debugPrint
logging at configure() and setQueueFromTracks() time to show the
actual baseUrl + per-track resolved URL on the next reproduction.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Refactors the new describe block to use the existing setupFresh()
helper pattern (mirrors setup() but with empty profile/folder lists
so we can verify the test response is what populates the dropdowns),
the lidarrSection() scoping helper (the cover-providers and SMTP
panels also have "Save changes" buttons that match unscoped queries),
and the correct mockQuery({ data: ... }) shape used elsewhere in the
file. Drops the duplicate afterEach (the file already has a global
one at line 131).
No behavior change in the tests themselves — same three cases.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three new cases:
- Save on a fresh form calls test first, fills auto-defaults from the
response, then put-config with non-zero defaults
- Save aborts when test fails and surfaces the error code
- An operator-picked default value survives the test → save sequence
(auto-default $effect only fills when current value is 0/'')
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Click sequence on a fresh Lidarr config: enter URL + API key → click
Save. The page now calls testLidarrConnection first, stashes the
returned profiles/folders into local state, lets the existing
auto-default $effects pick first-of-each, then sends the put-config
request with non-zero defaults. If the test fails, the save aborts
and the error surfaces in saveError.
The explicit Test connection button keeps working and additionally
populates the dropdowns from the same response so an operator can
inspect/adjust before saving.
Each dropdown reads from its test-state array when present, falling
back to the existing query data (which only works after the first
successful save). list_errors entries surface as inline notes under
their respective dropdowns.
Closes the chicken-and-egg where saving required defaults that could
only be fetched after saving.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Mirrors the server-side extension of POST /api/admin/lidarr/test. The
integrations page consumes these in the next commit to pre-fill the
quality/metadata/root-folder dropdowns on first-time Lidarr setup.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Updates the happy-path test stub to handle the three list endpoints
Lidarr exposes (qualityprofile / metadataprofile / rootfolder) and
asserts profiles + folders make it through to the response.
- Adds a partial-failure case where one list endpoint 5xxs; verifies
ok=true, the failing list is omitted, and list_errors carries its
bucket code.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends POST /api/admin/lidarr/test response to include quality_profiles,
metadata_profiles, root_folders, and an optional list_errors map when the
Lidarr connection succeeds. The three list fetches run in parallel after
the ping; any per-list failure goes into list_errors so the response can
still report ok=true with whatever data did come back. Failed-ping
response shape is unchanged.
This lets /admin/integrations populate its dropdowns on a single
round-trip during first-time setup, fixing the chicken-and-egg where
the dropdowns previously gated on cfg.Enabled (which can't be true
until the first save, which itself needs non-zero defaults).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Cover-art Image.network calls were passing server-relative URLs
(/api/albums/<id>/cover) straight to NetworkImage, which interprets
"no scheme" as file:/// and crashes with "No host specified in URI".
Same root cause regardless of HTTPS or HTTP server.
ServerImage wraps Image.network with a Riverpod read of
serverUrlProvider and prefixes the configured base URL. Absolute URLs
(e.g. discover screen's Lidarr image_urls) pass through unchanged.
Three call sites updated: album_card, artist_card, playlists_list_screen.
discover_screen left as-is — its row.imageUrl is already absolute (Lidarr
returns full URLs from MusicBrainz / Spotify metadata) and it has a
meaningful errorBuilder that ServerImage doesn't expose.
Also adds a defensive check in audio_handler.setQueueFromTracks: if
the constructed stream URL ends up scheme-less, throw a StateError
naming baseUrl + track.streamUrl + track.id instead of letting it
fall through to ExoPlayer which surfaces a confusing "Cleartext HTTP
traffic to 127.0.0.1 not permitted" error (Android's URL parser
defaults a scheme-less URI to localhost). User reported this exact
confusing error against an HTTPS prod server; the better message
will pinpoint where the empty baseUrl comes from on next reproduction.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two related Android setup gaps caught when running on a clean Pixel 6
emulator:
1. MainActivity extended FlutterActivity, but the audio_service
plugin requires AudioServiceActivity. AudioService.init() threw
PlatformException("The Activity class declared in your
AndroidManifest.xml is wrong"), and the partial-engine-reinit that
followed cascaded into a flutter_cache_manager SQLite EXCLUSIVE
lock crash. Both errors clear with the correct base class.
2. The main manifest was missing two production permissions:
- POST_NOTIFICATIONS — Android 13+ runtime requirement for the
media notification (now-playing tile in the system tray).
audio_service auto-prompts at init() time once declared.
- ACCESS_NETWORK_STATE — used by ConnectionErrorBanner +
flutter_cache_manager for connectivity awareness.
Note for future contributors: src/main/AndroidManifest.xml IS the
production manifest. The src/debug and src/profile variants only add
dev-only entries (e.g. INTERNET re-declared in debug for hot-reload).
There is no separate "release" manifest in Flutter's convention.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The bootstrap-admin-from-env-vars flow was a holdover from before
self-registration could create the first admin. Now that handleRegister
+ CreateUserFirstAdminRace promotes the first user to register on an
empty users table (verified shipped in v2026.05.08.2 / #376), the
bootstrap path is just a second source of truth that confuses operators
(and leaves an "admin" account in the DB that nobody asked for).
Removes:
- internal/auth/bootstrap.go + its test
- The auth.Bootstrap call from cmd/minstrel/main.go
- AuthConfig + AdminBootstrapConfig structs from config
- MINSTREL_AUTH_ADMIN_USERNAME / _PASSWORD env reads + their test
- The auth: block from config.example.yaml
- Bootstrap-related comments in docker-compose.yml + README.md
The README quickstart now points operators at /register on first start
instead of "watch the logs for a one-time password."
Existing instances keep their bootstrap-admin row in the DB; operators
who want it gone can register a new admin via /register, promote them
in /admin/users, then delete the old bootstrap user (last-admin guard
will require the new admin to be promoted first). No migration needed.
Recovery story for forgotten admin passwords now hinges on Fable #321
(admin password reset CLI) — currently the only path back in if no
other admin exists.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CI caught six issues against the new admin slice:
- AsyncValue<T> in this Riverpod 3.3.1 codebase exposes `.value`
(returns T?), not `.valueOrNull`. Switched the three admin readers
to match the established convention (player_bar, queue_screen,
now_playing_screen all use `.value`).
- home_screen.dart still has ctx.push calls in _albumsRow / _artistsRow
(carousel tap handlers) — restored the go_router import that
Task 2 over-eagerly removed.
- admin_user_edit_sheet.dart had a single-line `if (!await ...) return;`
that violated curly_braces_in_flow_control_structures. Wrapped in
braces.
- admin_quarantine_item.dart doc comment had `<top>` placeholders that
the analyzer flagged as unintended HTML. Rephrased without angle
brackets.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Single-scroll layout with two _SectionHeader-anchored blocks. Users
rows tap into AdminUserEditSheet. Invites row shows token (monospace)
with copy button, optional note, and a redeemed badge if applicable.
"Generate invite" opens a small dialog with an optional note field
(server hardcodes 24h TTL — admins can't customise expiry). On
success the new token is shown in a copy-once result dialog.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
AdminUserRow shows username + admin/auto-approve badges; tap opens
AdminUserEditSheet bottom sheet.
Edit sheet: SwitchListTile pair for is_admin and auto_approve_requests
(both surface server errors via SnackBar, including the last-admin
guard). Reset password takes a typed-input dialog (>=8 chars enforced
client-side; admin supplies the new password). Delete routes through
TypedConfirmSheet ("DELETE") and pops the sheet on success.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends AdminUsersController (build-only since slice 2) with setAdmin,
setAutoApprove, delete, resetPassword. is_admin/auto_approve toggles
optimistically swap the row and roll back on server error (including
the last-admin guard 4xx). resetPassword is admin-supplied; no
auto-generation mode exists server-side.
Adds AdminInvitesController + adminInvitesProvider with create
(prepends new invite to the list, returns it for the copy-once
dialog) and revoke (optimistic remove + rollback).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ExpansionTile per aggregated quarantine row — collapsed shows track,
artist · album, and "{N} reports: {top reason}". Expanded reveals the
nested per-user reports with their reasons + optional notes.
Three-dot menu offers Resolve / Delete file / Delete via Lidarr; both
deletes route through TypedConfirmSheet which requires the user to
type "DELETE" before the action button enables.
Adds AdminQuarantineController with optimistic-removal + rollback,
mirroring AdminRequestsController's pattern.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
AdminRequestsScreen renders rows from adminRequestsProvider, joining
requester user_id → username via adminUsersProvider for display
(falls back to UUID prefix when the users list hasn't loaded yet).
Approve fires immediately; Reject confirms via dialog. Both are
optimistic with rollback in the controller.
Adds AdminUsersController build-only (the read side) so the row join
works; mutation methods land in Slice 3 alongside the Users screen.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
AsyncNotifier removes the row optimistically on approve/reject; rolls
back to the pre-mutation list on server error. Invalidates
adminCountsProvider so the landing badge stays accurate.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Settings gains an _AdminSection that renders an "Admin" tile (Shield
icon → /admin) only when the current profile has is_admin = true.
Non-admins see the section as SizedBox.shrink() so the const ListView
children stay const.
Provides a discoverable second entry to the admin landing alongside
the kebab overflow shortcut.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ListView of three AdminSectionCard tiles (Requests / Quarantine /
Users) showing live counts from adminCountsProvider. Pull-to-refresh
re-fans-out to the three list endpoints. Admin/Quarantine/Users
sub-routes will land in subsequent commits.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Single landing-screen provider fans out to requests/quarantine/users
list endpoints in parallel and returns counts for the section-card
badges. Per-section AsyncNotifier controllers land alongside their
screens in subsequent commits.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Four small Dio-backed API classes mirroring the server's actual
contract (verified against internal/api/admin_*.go):
- Requests/Quarantine return flat lists; Users/Invites are enveloped
({"users": [...]}, {"invites": [...]}) and unwrap here
- setAutoApprove sends {auto_approve: bool} — different from the
response field name auto_approve_requests
- resetPassword takes admin-supplied {password: string}, returns 204
- Invite create takes optional {note: string}; expiry is server-fixed
at 24h
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Hand-rolled to match the server's actual JSON shapes:
- AdminRequest carries user_id (UUID, not username); kind switches
which title field is the display name
- AdminQuarantineItem aggregates reports per track with reason_counts
and a nested reports[] list, mirroring adminQueueRowView
- AdminUser includes display_name + auto_approve_requests
- Invite uses invited_by UUID + note (server hardcodes 24h TTL)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds /admin, /admin/requests, /admin/quarantine, /admin/users routes to
the ShellRoute (so PlayerBar persists). Redirect closure refuses any
/admin path when user.isAdmin is false. Server-side RequireAdmin
middleware remains the actual authority — this is UX.
Screens land in subsequent commits; this intermediate state won't build
until task 8/9 lands the AdminLandingScreen and the missing siblings
are stubbed out by task 11/13/16.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Search keeps its conditional clear-text button and gains the shared
nav widget after it. Discover/Playlists/Settings get an actions row
they didn't have before. The kebab is now reachable from every
top-level screen, which is the prerequisite for the Admin entry.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
HomeScreen drops its 5 ad-hoc IconButton actions in favour of the shared
widget. LibraryScreen gains an actions row it didn't have before. Both
now consistent.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Shared AppBar actions for top-level screens — three primary icons
(Home/Library/Search, current screen suppressed) plus a kebab containing
Playlists/Discover/Settings and (only when isAdmin) Admin. Replaces the
ad-hoc per-screen IconButton lists; centralises admin-gating logic in
one place so /admin entry can't accidentally leak to non-admins.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The +layout.svelte auth guard only treated /login as public, so any
unauthenticated visit to /register bounced to /login?returnTo=%2Fregister.
The "Register" link on the login page therefore looped back to itself,
making the first-user-becomes-admin bootstrap path (handleRegister +
CreateUserFirstAdminRace) unreachable in production.
Extracted isPublicRoute() into web/src/lib/auth/publicRoutes.ts so the
public-route set has one source of truth and is unit-testable. Test file
explicitly comment-tags /register as a #376 regression guard.
Closes the last gap in user-mgmt umbrella #376.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- audio_handler: override skipToQueueItem(index) -> _player.seek(zero, index: i)
- player/queue_screen.dart: list of MediaItems with current track highlighted
(left accent border + Now Playing badge), tap-to-jump on non-current rows
- /queue route + queue_music icon on /now-playing AppBar
Reorder + remove deferred for v1.
- models/playlist.dart (Playlist, PlaylistTrack, PlaylistDetail)
- api/endpoints/playlists.dart (list with kind=user|system|all, get detail)
- playlists/playlists_provider.dart (Riverpod family providers)
- playlists/playlists_list_screen.dart (with system-variant pill)
- playlists/playlist_detail_screen.dart (header + Play button + rows)
- /playlists + /playlists/:id routes wired
- Home AppBar: queue_music icon next to Search
Tracks with track_id=null render greyed-out + struck-through (matches
web's PlaylistTrackRow behavior). Tap a row plays from that position;
top-level Play button plays the full playable subset from track 0.
Final task of PR2 DRY pass. Extracts the verbatim
`offset + items.length >= total` math repeated across 8
createInfiniteQuery sites into a single Page<T>-shape helper.
The task spec described a bare-array helper signature
(lastPage: T[], pageSize-based stop), but no call site in this
repo uses that shape — every paged endpoint returns a Page<T>
envelope. The helper is named pageGetNextPageParam and matches
the actual canonical shape so the 8 copies could collapse.
Migrated:
- likes.ts: 3 sites (TrackRef, AlbumRef, ArtistRef)
- albums.ts: 1 site (AlbumRef)
- queries.ts: 4 sites (artists + 3 search facets)
history.ts left alone — uses has_more/total, different shape.
W-T3 (617477b) consolidated toast rendering into a single <ToastHost />
mounted in +layout.svelte, but page-level tests render components without
the layout, so screen.getByTestId('toast') and role=status queries broke.
Mount ToastHost in beforeEach (auto-cleaned by svelteTesting()) and reset
the store with clearToast() in afterEach so toasts can't leak between
tests.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add audit.WriteOrLog: a one-line wrapper around Write that logs at
Warn and swallows the error, matching the package contract that
audit failures must not break user-facing operations.
Migrate the 13 call sites across 7 files in internal/api/ from the
3-line "if err != nil { logger.Warn(...) }" shape to a single call.
audit.Write stays exported for tests + any future caller that
needs strict semantics.
Adds three tests: success (no log), failure-via-closed-pool (Warn
record with action+err keys), and nil-logger (no panic). Tests
skip when MINSTREL_TEST_DATABASE_URL is unset, matching the
existing harness convention.
The PR1-T2 admin handler migration replaced bespoke 500-class messages
("lookup failed", "refetch failed", "trigger failed", "bump failed",
"remove failed", "update failed", "schedule row missing", "re-read
failed", "read failed") with apierror.Internal(err), which forces the
wire Message to "internal server error". That violates the PR1
contract that errEnvelope{Code, Message} must remain byte-identical
for existing clients.
Add apierror.InternalMsg(message, cause) — a 500-class constructor
that preserves the call site's user-facing message while keeping the
cause attached for logging and errors.Is. Re-migrate the 12 admin
sites whose pre-1cc7eb6 source had a non-empty bespoke Message so
they restore the original wire string. Sites whose original Message
was empty stay on Internal(err) (humanization to "internal server
error" is a tolerable improvement, not a regression).
admin_smtp.go's send_failed site (line 129) was already preserved
via a raw &apierror.Error literal in 1cc7eb6 and needs no fix.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Rewrite writeErr(w, err) to wrap *apierror.Error via apierror.From,
preserving the existing {"error": {"code", "message"}} wire envelope.
Add writeErrWithLog helper for 500-class errors that need an operator
log line. Migrate all 13 admin_*.go handler files (~76 call sites) to
the new signature; T3 will sweep the remaining api package.
The old 4-arg writeErr is removed, so non-admin call sites in
internal/api will not compile until T3 lands. This is by design — T2
and T3 are paired.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Wires the auto_approve_requests user flag into the request creation
flow. When the flag is set (admin enabled it via /admin/users from
U2-T3), POST /api/requests transitions the just-created pending row
through Service.Approve inline, which dispatches to Lidarr the same
way a manual admin approve does.
Failures inside Approve (ErrLidarrDisabled, ErrDefaultsIncomplete,
network/Lidarr errors) leave the row pending — same fallback as
manual approve hitting the same path. The user gets a 201 either
way; admin can still resolve the row in /admin/requests if it
stayed pending. The handler logs the auto-approve failure with
user_id + request_id for observability.
The "actor" id passed to Approve is the user's own ID — they're
acting under the privilege the admin granted them via the toggle.
The trail of "admin set the flag" already lives in audit_log
(ActionAutoApproveToggle from U2-T2). Adding a per-auto-approval
audit entry is a future enhancement; for v1 the toggle audit plus
the request row's status transition is enough.
Test verifies the Lidarr-disabled fallback contract: a user with
auto_approve=true and Lidarr unavailable still gets 201 Created
with status=pending (no crash, no 500). The "auto-approve actually
succeeds" test path requires a Lidarr stub; deferred until a
broader Lidarr test fixture lands.
Cleans up the 8 svelte-check warnings I deferred in prior CI runs.
Two distinct categories — handling them differently:
REAL REACTIVITY BUGS — fixed:
- AlbumMenu.svelte / ArtistMenu.svelte: cachedTracks initialized
from the `tracks` prop only at component creation. When the menu
is reused across rows (operator opens menu on one album, then
another), the cache from the first open serves stale tracks.
Fix: initialize cachedTracks to null; ensureTracks() reads the
current `tracks` prop on each call before falling through to
the network fetch.
INTENTIONAL PATTERNS LINT FLAGGED — suppressed with explanatory
comments:
- AlbumMenu / ArtistMenu / PlaylistCard menu containers: the
onclick={(e) => e.stopPropagation()} is a guard against the
window-level "any click closes the menu" listener — not
user-facing interaction. A paired keyboard handler is
intentionally absent (Escape-to-close lives on svelte:window
onkeydown). Suppressed a11y_click_events_have_key_events with
explanatory comments.
- CompactTrackCard overlay div: catches clicks on its inner
LikeButton + queue button so they don't bubble to the wrapping
play-card button. Inner buttons carry their own keyboard
handling. Suppressed both a11y_click_events_have_key_events and
a11y_no_static_element_interactions.
- theme.svelte.ts module-scope `_resolved = $state(resolve(_theme))`:
module-init reads _theme to seed _resolved; subsequent updates
go through setTheme() and the matchMedia listener. $derived
doesn't apply at module scope. Suppressed state_referenced_locally
with an explanatory comment.
Promised in the prior fix-forward: "I'll fix bugs when I see them
or open a follow-up task with a specific name, not park them in an
umbrella." Following through.
Two real algorithm bugs in F-T1's For-You composition + Discover
allocator. Both surfaced as failing unit tests under go test -race.
1. redistributeSlots was re-redistributing a bucket's full deficit
on every pass instead of just the residual. The loop computed
`deficit = b.want - final[i]` each iteration, but final[i] for
a deficit bucket never increases (its supply is exhausted), so
pass N saw the same deficit as pass N-1 and kept shoveling it
to peers. For [want:40 avail:100, want:30 avail:0, want:30 avail:100],
four passes pushed cross-user's deficit into dormant+random four
times each, hitting the 100-slot clamp at the end and producing
[50, 0, 50] instead of the spec'd [55, 0, 45].
Fix: track per-source `redistributed[i]` and subtract it from the
deficit each pass. Multi-pass behavior still works for the case
where a peer's supply runs out mid-distribution.
2. tieBreakHash used FNV-1a 64-bit with trackID + dateStr appended.
For dateStrs differing only in the last character ("2026-05-07"
vs "2026-05-08"), the FNV state diverged only in low bits at the
final byte; multiplication by FNV_prime propagates upward but the
relative ordering of 60 small candidate UUIDs (which differ only
in their last byte) ended up identical across the two dates. The
For-You head/tail test asserted that the tail's first 5 should
change across days; it didn't.
Fix: switch to SHA-256 truncated to 8 bytes. SHA-256 has full
avalanche, so any single-bit input change roughly half-flips the
output bits and meaningfully reorders.
The hash isn't security-load-bearing; we just need strong avalanche
for tiny dateStr deltas. Determinism (same inputs → same output) is
preserved.
Three Go lint hits + two web test failures, all from the U3 push.
- internal/mailer/mailer.go: SentEmail.HtmlBody → HTMLBody (Go's
initialism convention; revive flagged); FakeSender.Send unused
ctx parameter renamed to _.
- internal/audit/audit_test.go: removed dead validUUID helper. It
was added speculatively in U1-T1 and never called by any test.
pgtype import stays — other tests use it.
- web/src/routes/settings/settings.test.ts: existing ListenBrainz
tests queried `getByRole('button', { name: /save/i })` which
worked when the page had only one Save button. The new Profile
card adds a "Save profile" button that also matches the regex,
triggering "found multiple elements". Anchored to /^save$/i for
exact-match. The newer Save profile tests still use
/save profile/i which is unique.
This is the same shape of test-fixture-lag I owe an answer for: I
landed new content that broke an existing test, and the existing
test had a too-loose selector. The right fix is to tighten the old
selector now (this commit) and to flag this pattern — selectors
that regex-match by partial words — as a candidate for the DRY
pass / test-utils consolidation.
Five fixes in one commit, all from the U1+U2+U3 push:
1. internal/api/library_test.go's Mount call wasn't updated when
U3-T4 added the mailer.Sender param. 17 args, sig wants 18.
Adds a trailing nil for the mailer. Same shape of test-fixture
lag the project has hit before — flagged as a recurring pattern
for the DRY-pass slice.
2. web/src/routes/settings/settings.test.ts mocked $lib/api/me with
`getAPIToken: vi.fn()` (no return value). The new /settings API
Token card's $effect calls `getAPIToken().then(...)` which threw
"Cannot read properties of undefined (reading 'then')" on every
ListenBrainz test. Default the mock to mockResolvedValue so any
test that doesn't override gets a valid resolution.
3. web/src/routes/admin/users/users.test.ts queried row buttons by
/^Delete$/ but T3 added per-user aria-labels ("Delete alice"),
so the row buttons' accessible name is no longer "Delete". The
modal's confirm button has no aria-label so its name IS "Delete"
exactly. Tests now click by per-user aria-label first, then by
/^Delete$/ for the modal confirm.
4. Same file: /Set password/i regex matched both the dialog's
submit button AND the row's "Reset password for ..." buttons
(because regex `Set` matches "ReSet" case-insensitively). Switched
to /^Set password$/ exact-match.
5. web/src/routes/reset-password/[token]/reset-password.test.ts had
/New password/i which matched both the "New password" and
"Confirm new password" labels. Switched to exact-match string
selectors.
6. web/src/routes/forgot-password/+page.svelte's onSubmit handler
had no catch — a rejected forgotPassword() bubbled as an
unhandled rejection in vitest. Added a silent catch since the
page intentionally shows the same success message regardless of
outcome (no-enumeration posture).
Three CI fixes from the U1+U2+U3 push:
- internal/api/admin_users.go imported the standalone
github.com/jackc/pgconn — that module isn't in go.sum and
go vet caught it. Switch to github.com/jackc/pgx/v5/pgconn
(the path used elsewhere in the package, e.g. auth_register.go
and me_profile.go).
- /reset-password/[token] dereferenced page.params.token without
the undefined narrowing svelte-kit's typegen requires. Coalesce
to '' on read and reject empty token at submit time with a
clear error.
- /admin/integrations SMTP card had a label without an associated
control as the TLS row's leading column header. Switched to a
span — the actual checkbox lives in the wrapping label below it.
Completes the U3 frontend that T5a's crashed dispatch left half-done.
- /reset-password/[token] — public; new password + confirm. Calls
POST /api/auth/reset-password with the URL's token. invalid_token,
password_too_short, and mismatched-passwords surface as inline
errors. Success redirects to /login?reset=ok.
- /admin/integrations gains an SMTP card (alongside Lidarr) with
enabled toggle + all fields + Save and Send-test-email buttons.
Password input is a separate state from the loaded form, so empty
submission preserves the stored server-side value (matches the
backend's empty-password-preserves contract). Test button error
codes (no_email_on_file / not_configured / send_failed) surface
as actionable toasts.
Tests cover both new pages plus extensions to settings + integrations
test files for the cards landed in T5a.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Lands four of the U3 frontend surfaces. Splits T5 because the
dispatch covering everything in one shot crashed the runtime
mid-execution; this is the salvageable half.
- /settings gains three cards alongside Appearance: Profile
(display name + email), Password (current + new + confirm with
client-side mismatch check), API Token (display + copy +
regenerate-with-double-click-confirm). Server error codes map
to clear toasts.
- /forgot-password — public; takes an email and always shows the
success message regardless of whether the email is on file
(mirrors the server's no-enumeration posture).
- Login page gets a "Forgot password?" link below the existing
register link.
- API client functions for the four /me endpoints (changePassword,
updateProfile, getAPIToken, regenerateAPIToken), the SMTP admin
trio (getSMTPConfig, updateSMTPConfig, testSMTPConfig), and the
forgot/reset auth pair (forgotPassword, resetPassword).
Tests for these surfaces + /reset-password page + admin SMTP card
land in T5b (next follow-up).
POST /api/auth/forgot-password and POST /api/auth/reset-password.
Forgot-password ALWAYS returns 200 with empty JSON to prevent
enumeration of registered emails. Side effect: when email matches
a user with email-on-file, generates a 32-byte hex token (24h
TTL), inserts into password_resets, and sends the reset email via
the mailer. Mailer failures are logged (not surfaced) and the
audit log carries metadata.email_match for operator visibility.
Reset-password atomically claims the token via UsePasswordReset
(:execrows; concurrent calls can't both succeed). On rows=1,
hashes the new password and writes via ChangeUserPassword.
Returns 204 on success, 400 invalid_token on stale/used/missing
tokens, 400 password_too_short for short passwords. Audits
ActionPasswordResetByEmail.
Wires the mailer.Sender into the handlers struct via Mount;
production sender (NewSMTPSender) constructed in server.Router();
tests inject FakeSender via testHandlers default. The reset
URL embedded in the email is derived from r.Host (no PublicURL
config setting in v1; self-hosted operators see their own
hostname).
Tests cover happy-path send + token-row insertion, unknown email
returns 200 with no send, mailer failure still returns 200, reset
happy path verifies bcrypt match + used_at set, already-used
token 400, expired token 400, short password 400, bogus token 400.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
New internal/mailer/ package:
- Sender interface with two impls: SMTPSender (production, reads
smtp_config at send time so admin edits apply without restart;
uses stdlib net/smtp + STARTTLS) and FakeSender (test/concurrent-
safe call recorder).
- Embedded text + HTML templates for the password reset email,
rendered via stdlib text/template + html/template. The HTML
uses the FabledSword forest-teal accent color.
- ErrNotConfigured surfaces when smtp_config.enabled is false or
required fields are empty; callers like the future forgot-password
handler will treat this as "log and pretend success" to avoid
user-enumeration leaks.
Three admin endpoints under RequireAdmin:
- GET /api/admin/smtp-config — returns the singleton; password
field is masked ("***" or "").
- PUT /api/admin/smtp-config — updates settings. Validates
host + from_address are non-empty when enabled=true. Empty
password in the request preserves the stored value (so the
operator doesn't have to re-enter it on every save).
- POST /api/admin/smtp-config/test — sends a real test email to
the calling admin's email. 400 if admin has no email; 500 with
the error message on send failure (so the operator can debug
config without grep-then-trace through logs).
Tests cover the password-mask, password-preservation-on-empty,
enabled-requires-host-and-from validation, and the no-email-on-file
rejection. Mailer unit tests cover the fake recorder and template
rendering. The "real SMTP send" path needs a live server and isn't
covered in CI; the FakeSender covers that role for downstream tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Four authenticated endpoints for the user's own account:
- PUT /api/me/password — change own password. Caller must
supply current_password (verified via bcrypt). Distinct from
admin-driven reset (which doesn't require knowing the old).
Audits ActionPasswordChangeSelf.
- PUT /api/me/profile — set display_name + email. Both fields
are nullable; empty string clears, omitted leaves unchanged.
Email is lowercased before store + format-validated. Unique
violation → 409 email_taken.
- GET /api/me/api-token — returns current API token (for
copy-paste into Subsonic clients).
- POST /api/me/api-token — regenerates token. Old one stops
working immediately. Audits ActionTokenRegenerate.
All four use the existing RequireUser middleware on the authed
sub-router; audit writes are best-effort (logged on failure).
Tests cover happy paths, wrong-password 401, password-too-short
400, email-invalid 400, email-taken 409, clear-by-empty-string,
omit-leaves-unchanged, token GET + regen.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds users.email (optional, lowercase-unique via partial index),
smtp_config singleton, and password_resets tokens. Plus the
queries U3-T2 / T3 / T4 use:
- ChangeUserPassword: self-service password change. HTTP layer
verifies the current password before this fires.
- UpdateUserProfile: set display_name + email together.
- RegenerateApiToken: invalidate old API token.
- GetUserByEmail: case-insensitive lookup for forgot-password.
- GetSMTPConfig + UpdateSMTPConfig: admin SMTP settings CRUD.
- CreatePasswordReset / GetPasswordReset / UsePasswordReset
(:execrows for atomic claim) / DeleteExpiredPasswordResets
(cron-style cleanup, not yet wired).
Email column is nullable; users without email have admin-reset
as their only password-recovery path. The lower() unique index
allows many NULLs and prevents case-insensitive duplicates.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Extends the U1 /admin/users page with the four U2 admin endpoints.
- "New user" button opens a modal: username, optional display
name, password + confirm, optional admin checkbox. Validates
password match client-side; surface server-side errors
(username_taken, username_invalid, password_too_short).
- Per-row "Delete" opens a confirm modal explaining the cascade
(plays, likes, sessions). Last-admin guard surfaces as a clear
toast if the server refuses.
- Per-row "Reset password" opens a small modal: new password +
confirm. Toast confirms success.
- Per-row "Enable / Disable auto-approve" toggles the
per-user flag (the #355 sub-feature surface). Inline button
state reflects the current value.
AdminUser type extended with auto_approve_requests; the badge
appears next to the admin badge when enabled.
Tests cover create-user submit, delete confirm flow, last-admin
toast on delete, reset-password submit, auto-approve toggle.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Four new admin endpoints under existing RequireAdmin middleware:
- POST /api/admin/users — admin-creates-user. Body
{username, password, display_name?, is_admin?}. Same username
+ password validation as the public /register handler. 409
on duplicate. Audits ActionCreateUserAdmin.
- DELETE /api/admin/users/{id} — hard delete. Last-admin guard
refuses delete when target is the only admin (409). Schema's
ON DELETE CASCADE on user-FK tables handles plays/likes/
sessions cleanup. Audits ActionDeleteUser with target's
username + was_admin flag.
- POST /api/admin/users/{id}/reset-password — body
{password}. 8-char minimum. Admin sets a new password
without knowing the old one. Audits ActionPasswordResetAdmin.
- PUT /api/admin/users/{id}/auto-approve — body
{auto_approve: bool}. Toggles the per-user flag added in T1
(the #355 sub-feature surface). Audits ActionAutoApproveToggle.
adminUserView shape extended with auto_approve_requests so the
list and toggle responses carry the flag. ListUsers SQL query
updated to include auto_approve_requests; generated Go updated
to match. Tests cover happy paths, last-admin guard on delete,
password validation, duplicate username, and the auto-approve
round-trip.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds users.auto_approve_requests boolean default false (the #355
sub-feature surface; the request-flow handler that honors this
flag is U2.5 follow-up work).
Four new sqlc queries for the U2 admin endpoints:
- CreateUserAdmin: admin-driven user creation, accepts all five
fields explicitly.
- DeleteUser: hard delete; schema's ON DELETE CASCADE foreign
keys handle plays/likes/sessions cleanup. Last-admin guard
lives in the HTTP layer (next task).
- ResetUserPassword: admin sets a new hashed password without
knowing the old one.
- UpdateUserAutoApprove: toggles the new boolean.
Self-service equivalents (knows-current-password change, etc.)
are U3 work and use distinct queries.
New admin route /admin/users with two sections:
- Accounts — table of all users (username, optional display name,
admin badge, created_at). Per-row "Make admin" / "Remove admin"
button calls PUT /api/admin/users/{id}/admin. Last-admin guard
surfaces as a toast ("Can't remove the last admin — promote
someone else first").
- Invites — list of active invites with Copy + Revoke per row.
"Generate invite" button creates a 24h token. Operator shares
the token with the invitee, who enters it on /register.
AdminTabs gains a "Users" tab (5 tabs total). New typed client
functions in admin.ts (listUsers, updateUserAdmin, listInvites,
createInvite, deleteInvite) plus query factory functions
createAdminUsersQuery / createAdminInvitesQuery and query keys
qk.adminUsers / qk.adminInvites.
Tests cover: list rendering, admin badge, promote/demote actions,
last-admin guard toast, invite generation, revoke flow, empty states.
AdminTabs.test.ts updated to assert 5 tabs and Users active state.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Public /register form: username, optional display name, password +
confirm, optional invite token. Submits to POST /api/auth/register
via the new register() helper in auth/store.svelte.ts. On success
the server sets the session cookie and the frontend redirects to /;
on error the page shows a code-specific message (invite_required,
invite_invalid, username_taken, etc.).
Login page gets a "Don't have an account? Register" link below the
form for symmetry.
Tests cover form rendering, password-mismatch client-side rejection,
the happy-path submit + redirect, and the invite-invalid error
mapping.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Five admin endpoints under existing RequireAdmin middleware:
- GET /api/admin/invites — list active + recently-redeemed
- POST /api/admin/invites — generate 24h invite, returns
{token, expires_at, ...}. Audits ActionInviteCreate.
- DELETE /api/admin/invites/{token} — revoke unredeemed invite.
Audits ActionInviteRevoke.
- GET /api/admin/users — list all users (id, username,
display_name, is_admin, created_at).
- PUT /api/admin/users/{id}/admin — toggle is_admin with
last-admin guard. Audits ActionPromoteAdmin / ActionDemoteAdmin.
Last-admin guard counts admins, refuses demotion of the sole
admin with 409 'last_admin'. Race window between count and update
is acceptable for v1 — worst case is 'no admins left,' which the
env-driven bootstrap or CLI reset can recover from. Common path
('admin demotes themselves') is now blocked.
Adds ListUsers, CountAdmins, UpdateUserAdmin sqlc queries to
users.sql.
Tests cover: invite create/list/delete round-trip, non-admin gets
403, user list, promote happy path, last-admin demotion refused,
two-admins demotion allowed.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
POST /api/auth/register accepts {username, password, invite_token?,
display_name?} and creates a user. Race-safe first-admin path: when
the users table is empty, the SQL query (CreateUserFirstAdminRace)
inserts is_admin computed from a SELECT NOT EXISTS subquery — no
serializable isolation needed; concurrent empty-state registrations
both end up admin (benign).
Past first-admin, registration_settings.mode dictates: 'invite_only'
(default) requires a valid unredeemed unexpired invite token, which
is atomically claimed via RedeemInvite (rows-affected returns from
sqlc's :execrows directive). 'open' mode skips the invite check.
On success: hashes password (bcrypt), mints session token + cookie
matching handleLogin's shape, mints a separate api_token for
Subsonic clients, audits ActionRegister + ActionInviteRedeem
(best-effort — a failed audit write does NOT fail the user-facing
operation).
Validation: 3-32 char usernames (alphanumeric + underscore +
hyphen), 8-char minimum password. Duplicate username surfaces as
409.
Tests cover: first-user-becomes-admin, invite-only requires token,
valid token redeems + non-admin role, invalid token 400, open mode
skips check, duplicate username 409, password too short 400,
username format 400, race scenario asserts at-least-one-admin.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Schema for user management U1: display_name on users; user_invites;
registration_settings singleton (default 'invite_only'); audit_log.
Plus the internal/audit package centralizing the action-name
vocabulary and JSON metadata marshaling so handlers don't repeat
boilerplate.
Race-safe first-admin uses a query-shape primitive
(CreateUserFirstAdminRace's WHERE NOT EXISTS subquery) rather than
a schema-level constraint. Concurrent empty-state registrations
both see 'no users yet' and both insert as admin — fine, having
two admins from the start is benign; what matters is at-least-one.
users.username uniqueness arbitrates if the two callers picked the
same username.
CreateUser signature gains display_name (nullable); existing
bootstrap call sites pass nil. The audit package declares the full
U1+U2+U3 action vocabulary upfront so subsequent slices are purely
additive on the caller side.
Tests cover audit Write with + without metadata and that every
declared action constant persists correctly.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Click on the For-You tile's play button now triggers a synchronous
refresh, then enqueues + auto-plays the freshly-built playlist.
Two-click-target tile pattern: tile BODY click still navigates to
the playlist detail page (existing behavior); the play button is
the new generate-and-play action.
Behavior is gated on playlist.system_variant === 'for_you' — every
other playlist (user, Discover, Songs-like-X) keeps the existing
"fetch detail, enqueue, play" flow. The atomic-replace BuildSystem-
Playlists creates a new playlist_id, so the play handler uses the
refresh response's id for the follow-up getPlaylist call rather
than the stale tile prop's id.
Empty-library response (playlist_id: null) is a no-op — clicking
play in a degenerate-empty library doesn't error, it just doesn't
start playback. Caches are invalidated post-refresh so the home
view re-fetches and the tile re-renders with the new id, avoiding
the corner case of a click-then-tile-body 404.
Tests cover the refresh-and-play happy path, the empty-library
no-op, and that non-For-You playlists keep the existing flow.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
bvandeusenandClaude Opus 4.7