Go's //go:embed needs web/build/ to exist at compile time. CI now
runs 'npm ci && npm run build && npm test' ahead of the Go steps
so the embed directive sees real, freshly-built assets.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Web assets are built in the node stage, copied into the go stage before
go build so //go:embed picks them up, then the minimal slim runtime
carries only the final binary and ffmpeg.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- NotFound wrapper routes unknown /api/* and /rest/* to a JSON 404
- Everything else falls through to the embedded web handler, which
resolves static assets or returns index.html for SPA deep links
go:embed paths are relative to the source file and cannot reach parent
directories, so internal/web/embed.go could not see web/build/ without
duplicating the placeholder. Relocating to web/embed.go lets the
directive resolve to the real SvelteKit build output with no copy step.
- SvelteKit 2.x + Svelte 5 + TypeScript + Vite 5
- adapter-static with fallback:'index.html' for SPA deep-link routing
- Tailwind configured with dark-only palette matching spec
- Placeholder landing page at /
- Committed web/build/index.html placeholder (via .gitignore negation) so
go:embed works before the SvelteKit build has been run
CI's golangci-lint run flagged three files; two pre-existed this
branch but the Plan 3 seedTrackWithFile struct-literal alignment
is new. Applying gofmt across all three keeps the lint baseline clean.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Four-task TDD plan for /api/albums/{id}/cover and /api/tracks/{id}/stream:
scaffold media.go helpers (Task 1), cover handler (Task 2), stream handler
(Task 3), production Mount wiring + route-registration regression (Task 4).
Uses seedTrackWithFile helper to materialize real bytes on disk so Range /
If-Modified-Since tests hit http.ServeContent's real code path.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Self-review fixups: make "first track" explicit (by disc/track order) and
replace the awkward "so existing tests continue to compile" phrasing with a
clearer description of why newLibraryRouter needs extending.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Covers /api/albums/{id}/cover and /api/tracks/{id}/stream — the byte-serving
endpoints Plan 2's cover_url / stream_url forward references point at.
Duplicates protocol-agnostic helpers locally per subsonic-is-legacy direction
rather than extracting a shared package.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
12 TDD-style tasks covering SQL additions, DTOs, conversion helpers,
per-endpoint handlers (tracks, albums, artist detail, artist list, search),
route wiring, and an end-to-end smoke + PR. Follows the approved design at
docs/superpowers/specs/2026-04-20-web-ui-library-reads-design.md.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Plan 2 of the web UI rollout. Covers GET /api/artists (paged, two sort
modes), artist/album/track detail, and unified search with enveloped
pagination and forward-reference stream/cover URLs.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
sessionTokenFromHTTP now matches extractBearerToken's trimming
behavior. Without this, a bearer header with trailing whitespace
(e.g. a buggy client or proxy) authenticates via RequireUser but
hashes the padded value on logout, silently no-oping and leaving
the server-side session alive.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wires the native JSON surface alongside the existing /api/admin
and /rest/* routes. Subsonic compatibility remains untouched.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Returns the authenticated user (id/username/is_admin). Shape
matches UserView used in the login response so SPA stores stay
typed against one interface.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Deletes the session row keyed by the cookie/bearer token and
clears the cookie on the client. Best-effort DB delete — logout
still succeeds for the client if the row's already gone.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Verifies bcrypt password hash, mints a session token, stores its
sha256 in the sessions table, and returns the token in both the
response body (for Flutter) and an httpOnly/SameSite=Strict
cookie (for the web SPA).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Introduces internal/api with Mount(), the {error:{code,message}}
response shape, and stubbed login/logout/me so later tasks can
land one handler at a time without breaking the build.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Resolves /api/* callers from session cookie first, Authorization
bearer second. Touches last_seen on success for future active-
sessions UI. Adds GetUserByID query used by the middleware.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Shared primitives for /api/* auth: mint a url-safe opaque token,
hash it for storage, verify a bcrypt password hash.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Stores sha256(token) plus user_agent + last_seen_at so future
active-sessions UI doesn't need another migration.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
10-task plan for the first implementation slice of the web UI
scaffold: sessions table, /api/auth/{login,logout}, /api/me, and
the RequireUser middleware (cookie + bearer). Stops short of the
library read endpoints and the SvelteKit project, which become
their own follow-up plans.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
First pass of the web UI design doc from the M1 close-out brainstorm.
Covers the stack (SvelteKit static + Go embed), auth model
(cookie + bearer from one endpoint), /api/* surface, shell layout
(sidebar + bottom player), first-cut feature scope, and how later
web UI work threads into M2–M5 alongside each backend milestone
instead of landing as a post-hoc M6.
Also ignores the local .superpowers/ brainstorming companion cache.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Scrubbing/seeking in clients was a no-op because every track shipped
with duration_ms=0. Shell out to ffprobe (already in the image) per
file during scan and record the parsed duration. ffprobe failures are
warned + recorded as 0 so a single bad file doesn't sink the scan.
Tightened the incremental skip to also require duration_ms > 0 so
existing libraries get backfilled on the next rescan instead of
needing a wipe.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
bvandeusen