- Flag now detects Postgres SQLSTATE 23503 (foreign_key_violation) on
UpsertQuarantine and surfaces ErrTrackNotFound instead of a wrapped
generic. T8's handler can now return 404 track_not_found cleanly.
Constraint-name guard ('track' substring) keeps a future user_id FK from
mis-mapping to the same error.
- AdminQueueRow.AlbumTitle is now string (not *string). albums.title is
NOT NULL upstream; the empty-string-to-nil dance was answering a
nullability question that doesn't exist in the schema.
- Add TestFlag_NonexistentTrackReturnsErrTrackNotFound covering the new
branch.
- dbtest.ResetDB.dataTables now truncates lidarr_quarantine + lidarr_quarantine_actions
alongside the other M2-M4 data tables. Without this, M5b service tests would
inherit residual state across runs.
- DeleteTrackFile godoc spells out the post-file/pre-DB failure window
reconciles via the next library scan; the function is retry-safe by design,
not atomic.
- Decode errors in LookupArtist/AlbumByMBID now wrap ErrInvalidPayload so
callers can errors.Is the same way they do against the M5a methods.
- Tighten TestLookupAlbumByMBID_BadJSON to assert on ErrInvalidPayload.
- Add empty-mbid rejection tests for both methods.
- Add TestLookupArtistByMBID_EmptyArrayReturnsErrNotFound (only the album
variant had the test before — symmetric coverage now).
- Normalize the network-error test's api key to key123 for consistency
with the rest of delete_test.go.
- Composite PK (user_id, track_id) already serves WHERE user_id queries;
the secondary (user_id, created_at DESC) index just amplified writes.
- ListQuarantineForUser now selects only the joined fields the SPA card
actually renders (~10 columns) rather than embedding three full structs
(~35 columns); halves wire/DB bandwidth before consumers exist.
- max(q.created_at)::timestamptz cast emits pgtype.Timestamptz instead of
interface{} so handlers can read latest_at without a type-assert.
- Adds Service tests for ListPending/ListByStatus/ListForUser, the album-
and unknown-kind validation branches, and an Approve happy path with a
stub Lidarr server that exercises both default-snapshot and override-
snapshot persistence.
- Adds Reconciler tests for Run (short tick + cancel), the album-not-yet-
in-library no-op branch, and the track-kind 'parent album present but no
tracks yet' branch.
Combined coverage on internal/lidarr*, lidarrconfig, lidarrrequests now
86.5% (lidarrrequests 81.5%), meeting the per-package >=80% target in
spec section 8.
Pending/Approved/Completed/Rejected tabs (active count only). Override modal
collapsed by default; reject modal with notes textarea. Lidarr-unreachable
toast surfaces sync-Approve failures. Cancel buttons in both modals are
Pewter ghost; Reject/Confirm reject use Bronze (the design system reserves
Oxblood for irreversible actions like Disconnect).
- Trim whitespace before checking the typed-confirm value (paste-with-spaces
was silently keeping the button disabled).
- Wrap onConfirmDisconnect in try/catch and surface disconnect_failed inline
in the modal — operators on flaky networks now see why nothing happened.
- Invalidate quality-profile + root-folder caches alongside config on both
save and disconnect, so a base_url change refetches the lists immediately.
Connection management for Lidarr — base URL, API key, default quality
profile, default root folder. Empty api_key on save preserves the
masked-saved key per backend semantics. Disconnect requires typed
"DISCONNECT" confirmation. Includes a placeholder MusicBrainz overrides
section for future integrations.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Replaces the :global scoped-style workaround in AdminSidebar with a proper
accent.tint color tier in tailwind.config.js. Tailwind generates bg-accent-tint
deterministically; the global class no longer leaks from a single component.
Future consumers (e.g. /admin/requests tab counts) get the same utility.
Hard route gate in admin/+layout.ts redirects non-admins to / before
the layout (or any child) renders. AdminSidebar reads the active route
from $app/state and applies a 12% accent-tinted bg + 2px forest-teal
left strip on the active item. Overview landing shows pending-request
count and Lidarr connected/unset, each linking to its sub-page. Shell
nav exposes the Admin link only to is_admin users.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Album- and track-kind requests can have null album_title/track_title (the
type allows it). Previously aria-label rendered 'Cancel request for —' which
is meaningless to a screen reader. Fall back to 'this album by <artist>' /
'this track by <artist>' so the cancel/listen buttons stay disambiguable.
Renders the caller's Lidarr requests as rows with kind pill,
StatusPill, and per-status actions: Cancel on pending (which
calls cancelRequest then invalidates qk.myRequests()), Listen
link on completed (deepest match wins: track > album > artist),
admin notes on rejected. Empty state uses the voice-rule
"Nothing requested yet." copy. Shell nav gains /requests
between /discover and /playlists, visible to all authed users
since the view is per-user.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Local-input + 250ms debounce drives Lidarr search; tab switch refetches
with new kind. Track-kind Request opens a confirm modal explaining the
album that will be added; Confirm fires createRequest, Cancel is a no-op.
Successful requests flip the card to 'requested' optimistically.
Shell nav now exposes /discover between Search and Playlists.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Single-prop component mapping LidarrRequestStatus to voice-rule labels and
semantic tones (warning/info/success/error). Used by /requests and /admin/requests
in subsequent tasks.
aria-labels now include the card title on all three button variants so
SR users navigating a grid can tell which card a button belongs to:
- "Request <title>" / "<title> is already in library" / "<title>
already requested"
The Kept pill gets role="status" so SR users hear the badge when it
appears (without aria-live's announce-on-mount noise).
The reserved-slot CSS (.badge-row { min-height: 22px } and
.actions { margin-top: auto }) was already in the scoped <style>
block; we drop the duplicate inline style="" attributes that existed
purely to satisfy jsdom's getComputedStyle. Tried stylesheet
introspection (document.styleSheets) as a replacement assertion, but
vitest's @testing-library/svelte renderer doesn't inject the scoped
<style> tag into jsdom (styleSheets.length === 0), so the two CSS
assertions are dropped with an explanatory comment. The layout
discipline is enforced visually at the consumer page rather than as
a "CSS exists in CSS" unit test.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
T14 of the M5a Lidarr plan. The Discover grid in M5a renders mixed
requestable / kept / requested cards from Lidarr search. Without
layout discipline the cards in a row land their titles at different
Y coordinates whenever the badge presence varies, which reads as
visual noise. DiscoverResultCard enforces:
- Flex column with `margin-top: auto` on `.actions`, so the action
button is anchored to the bottom of the card body regardless of
title/subtitle wrap differences.
- `.badge-row` always rendered with `min-height: 22px`, so the
title baseline holds even when no Kept pill is present.
Both load-bearing CSS values use inline style attributes — jsdom's
getComputedStyle does not resolve scoped <style> blocks, so the tests
verify positioning via inline styles directly. The remaining decorative
CSS (kept-pill background = accent at 15%, flex-direction, gap) lives
in a scoped <style> block.
State -> action mapping (sentence case per FabledSword voice):
requestable -> bg-action-primary "Request" with Plus icon
kept -> disabled ghost "In library" + Kept pill
requested -> disabled ghost "Requested"
Cover art falls back to a Lucide glyph (Disc3 / Album / Music2) when
imageUrl is absent. Adds lucide-svelte@^1.0.1 dependency.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- qk.adminRequests defaults to status='all' when no filter is set; the unfiltered
endpoint returns every status, so caching it as 'pending' was a latent mismatch.
- 60s staleTime on lidarrConfig + myRequests — both are session-stable enough
that refetching on every consumer-page mount produced needless flicker.
T13 of the M5a Lidarr plan. Three new client modules wrap api.get/post/put/del
helpers and the existing TanStack Query qk namespace:
lidarr.ts - searchLidarr() + createLidarrSearchQuery()
requests.ts - createRequest, listMyRequests, getRequest, cancelRequest;
createMyRequestsQuery(); cancel goes through apiFetch
directly because the backend returns the cancelled row body
(api.del's return type is fixed to null).
admin.ts - getLidarrConfig, putLidarrConfig, testLidarrConnection,
listQualityProfiles, listRootFolders, listAdminRequests,
approveRequest, rejectRequest; query factories for each
read; quality profiles + root folders take an enabled prop
so the call site decides when Lidarr is configured.
Shared LidarrRequestStatus / LidarrRequestKind enums and request/config/
search-result shapes added to types.ts. Per-module helpers (CreateRequestParams)
stay in their module files. testLidarrConnection returns a discriminated union
({ok:true,version} | {ok:false,error}) and never throws on ok:false so the
SPA can render either branch.
qk extended with lidarrSearch, myRequests, lidarrConfig,
lidarrQualityProfiles, lidarrRootFolders, adminRequests.
Tests mirror likes.test.ts (vi.mock('./client')) and cover URL construction,
query-param encoding, body shapes, the not-ok testLidarrConnection branch,
and qk additions. 32 new tests, 217 total passing.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Removes duplication between fabledsword-tokens.css and tailwind.config.js;
font stacks now live in one place so per-app overrides reach Tailwind utilities.
Add the canonical FS token palette (surfaces, text, action, semantic,
accent, radii, fonts) as CSS custom properties under web/src/lib/styles/
fabledsword-tokens.css, with a [data-fs-app="minstrel"] hook reserved
for future per-app accent overrides. Wire the tokens through Tailwind by
aliasing semantic colour utilities (bg-background, bg-surface,
bg-surface-hover, text-text-primary/secondary/muted, border-border,
bg-action-primary/secondary/destructive, accent, warning/error/info) to
the new variables, plus rounded-sm..xl and font-display/sans/mono. Load
Fraunces, Inter, and JetBrains Mono (400/500 only) via Google Fonts and
default the body to Inter. Existing components inherit the new palette
without per-page changes.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three admin-gated handlers: GET /admin/requests (list by status),
POST /admin/requests/:id/approve (with optional overrides), and
POST /admin/requests/:id/reject. testHandlers updated to inject a
real clientFn so Approve exercises Lidarr in integration tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Five admin-only handlers under RequireAdmin middleware: GET/PUT config
(api_key masked, empty key on PUT preserves saved), POST test (always
200, maps Lidarr errors to stable codes), GET quality-profiles, GET
root-folders. 10 HTTP integration tests, all green.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implements POST/GET/GET:id/DELETE /api/requests handlers delegating to
lidarrrequests.Service; wires routes in RequireUser group and threads
the service through Mount and server.go.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implements GET /api/lidarr/search?q=&kind=artist|album|track. Validates
kind and q, loads lidarr_config per-request, calls the matching Lidarr
Lookup* method, enriches each result with in_library (EXISTS by MBID
against artists/albums/tracks) and requested (HasNonTerminalRequestForMBID),
and returns a normalized JSON array. Adds lidarrCfg field to handlers struct
and threads lidarrconfig.Service through Mount and server.Router.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replaces the old X-API-Token-based RequireAdmin in middleware.go with a
context-aware RequireAdmin() that runs after RequireUser, checks
user.IsAdmin, and returns 403 {"error":"not_authorized"} for non-admins
or 500 {"error":"internal_error"} if RequireUser was bypassed. Updates
server.go to mount RequireUser then RequireAdmin on the /api/admin group.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add ArtistMBID/AlbumMBID to LookupResult so handlers can build the
full lidarr_requests row from album/track lookups (parent MBID is
needed for the AddAlbum API call and for the schema's NOT NULL
lidarr_artist_mbid)
- Trim trailing slash on BaseURL before joining paths to avoid
double-slash URLs when operators enter "http://lidarr.lan/"
- Check json.Marshal errors in AddArtist/AddAlbum
- Extract lidarrImage as a named unexported type instead of repeating
the anonymous struct three times
- Add NewClient(baseURL, apiKey) constructor with 30s default timeout
- Add ErrLookupFailed test coverage for the 4xx-non-auth branch on
both get and post helpers
- Rename TestListRootFolders_BadJSON -> _NullBody (it tests null,
not malformed); add a real malformed-JSON test
- Defensive separator guard on LookupAlbum.Secondary so an empty
ArtistName doesn't produce a leading " · "
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- HasNonTerminalRequestForMBID: use named parameter @mbid so the
generated Go signature is `mbid string` instead of the misleading
`lidarrArtistMbid string` (the value applies to all three MBID
columns, not just the artist column)
- CancelLidarrRequest: comment dual role of $2 (ownership guard +
decided_by audit field) for future readers
- CompleteLidarrRequest: document the per-kind one-of-three contract
for matched_*_id parameters
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
M5a is the foundation slice of M5 (Lidarr integration + quarantine).
Decomposed into M5a/M5b/M5c during brainstorming on 2026-04-29:
- M5a (this spec) — Lidarr connection + search/add + admin shell
- M5b — Quarantine workflow (per-user soft-hide, admin resolution)
- M5c — Radio suggested-additions (out-of-library MBIDs surfaced)
Each ships as its own PR with its own brainstorm/spec/plan cycle.
Key decisions captured:
- Permissions: search-all, add-admin via request queue
- Config: DB-only, Settings UI is the only entry point (no YAML)
- Settings shape: dedicated /admin/* route group, hard route gate
- Search UX: standalone /discover route, all three granularities
- Lifecycle: library scan as source of truth + 5-min reconciler worker
- Quality profile/root folder: default + per-add admin override
- UI lands at FabledSword design system bar (forest-teal accent;
Moss/Bronze/Oxblood action buttons; per project_design_system memory)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Companion to the dbtest.ResetDB change. The bootstrap integration
test fundamentally needs an empty users table — it exercises the
"first time admin is created" path, which Bootstrap() guards with
a count==0 check. So this test alone still has to TRUNCATE users.
To keep the operator's admin login intact on a shared dev DB:
- Before TRUNCATE, save every user that doesn't start with 'test-'.
- Use cfg.Username = 'test-admin' for the bootstrap call so the
test row is itself test-prefixed and gets cleaned up by other
tests' ResetDB calls.
- t.Cleanup restores the saved rows after assertions complete,
so admin/admin (or whatever password the operator set) keeps
working between test runs.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Integration tests previously TRUNCATEd the users table at setup,
which wiped the operator's admin login every time the suite ran
against a Postgres instance shared with their dev environment.
This commit:
- Adds internal/dbtest/reset.go exposing ResetDB(t, pool) and
TestUserPrefix. ResetDB truncates every data table EXCEPT users,
then deletes only users whose username starts with TestUserPrefix.
- Migrates 9 test files (subsonic/scrobble, subsonic/star,
recommendation/candidates, scrobble/queue, similarity/worker,
playevents/writer, playsessions/service, api/auth, api/me) to
call ResetDB instead of issuing TRUNCATE-with-users.
- Renames hardcoded test usernames to use TestUserPrefix so ResetDB
cleans them between runs. seedUser in api/auth_test now prefixes
internally; existing call sites pass bare names.
- Leaves auth/bootstrap_test.go alone — it legitimately tests
first-time admin bootstrap and must wipe users.
Verified locally: full integration suite with -p 1 runs cleanly
under the existing MINSTREL_TEST_DATABASE_URL setup, and the admin
row in the shared dev DB survives the run.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds _radioSeedId state and a module-level $effect.root that triggers a
/api/radio?seed_track=…&exclude=… fetch when queue consumption reaches
80%, appending new tracks without resetting the radio seed. Clears seed
on any non-radio enqueue (playQueue/enqueueTrack/enqueueTracks). Five
new vitest cases cover the trigger threshold, append logic, in-flight
guard, and manual-enqueue reset path.
Wire LoadCandidatesFromSimilarity as the primary candidate loader with
an ?exclude= query param for comma-separated UUID filtering; fall back
to LoadCandidates on error. Thread SimilarityWeight into ScoringWeights
and update testHandlers recCfg accordingly.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The combined seed_info CTE filtered the seed out when its genre was
NULL/empty. Restored the spec's two-CTE design (seed_artist + seed_tags)
so similar-artist matching works regardless of the seed's genre coverage.
Drops the genre workaround from the SimilarArtist test.
Third and final M4 sub-plan (Fable #347) — closes M4. Replaces M3's
whole-library candidate pool with a 5-way SQL UNION (LB-similar tracks /
tracks by similar artists / MB-tag overlap / likes-overlap / random
fill). Adds a new SimilarityScore × SimilarityWeight term to M3's Score()
formula. Web client auto-refreshes the queue when 80% consumed via a new
?exclude= query param. No lazy LB fetch (M4b's worker + random
augmentation handle sparse data). Fallback to M3 LoadCandidates on any
similarity-pool error.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Replace stub with full tickOnce: drains played tracks/artists via
ListPlayedTracksNeedingSimilarity / ListPlayedArtistsNeedingSimilarity,
calls LB SimilarRecordings/SimilarArtists, filters to local library via
bulk MBID lookup, enforces top-K=20 by score, and upserts similarity
rows. 429 aborts the entire tick; other errors log-and-skip. Ten
integration tests covering no-op, library filtering, top-K cap, 7-day
freshness cap, stale refresh, 429 abort, transient skip, and artist
pass.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Second M4 sub-plan (Fable #346). Periodic worker pulls track-track and
artist-artist similarity edges from LB's public /explore/* endpoints,
filters to the local library, stores top-20 per source track in two new
tables (track_similarity, artist_similarity). Hourly tick, batch=5,
weekly re-fetch cap per row, passive retry via timer. No auth (public
endpoints). Discovery within library handled by LB's collaborative-
filtering response naturally surfacing unplayed library tracks; spec
notes M4c will add a serendipity floor + lazy fetch + sparse-fallback.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds api.put helper, ListenBrainz API module with query/mutation
helpers, a /settings route with token + enable/disable UI, and 5
tests. Adds Settings to the Shell nav.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Exposes user ListenBrainz config (token_set bool, enabled, last_scrobbled_at)
via write-only token semantics; enforces no-enable-without-token at the API layer.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Bridges closed play_events to the scrobble_queue: loads the event,
applies the Qualifies() threshold, checks user LB config (enabled + token),
and inserts via ON CONFLICT DO NOTHING for idempotency. Best-effort —
callers (playevents.Writer hooks, Task 8) will log and swallow errors.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implements the HTTP client for ListenBrainz submit-listens with typed
sentinel errors (ErrAuth, ErrPermanent, ErrTransient, *RetryAfterError)
so the scrobble worker can branch on response semantics. All 9 httptest-
driven tests pass.
First M4 sub-plan (Fable #345). Outbound scrobble worker with batching +
exponential-backoff retry. Per-user token, plaintext storage, pull-based
30s timer worker, patient backoff (1m → 5m → 30m → 2h → 6h, 5 attempts).
New scrobble_queue table; sent rows deleted (canonical record stays in
play_events.scrobbled_at). Minimal /settings page in M4a as scaffold for
M6's full Settings sub-plan to extend.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Forgejo Actions fired both `push` and `pull_request` events for the same
commit when pushing to dev with an open PR — doubled CI runs on every PR
commit (e.g. #182 and #183 on PR #24's HEAD `95d68e3`). Drop the push
trigger entirely; PRs against main are the only quality gate that
matters in this workflow (dev → PR → main, never direct push to main).
Trade-off: direct pushes to dev with no open PR no longer get CI on
their own. Acceptable — the moment the PR opens, the pull_request event
fires and runs the same workflow.
Two carry-overs from M3 verification, bundled with the search-input
fix already on dev (b7a59a9).
1. Genre splitting in BuildSessionVector
The library has many tracks whose genre tag is a denormalized
multi-genre string ("Indie Pop; Pop; Alternative Pop"). Reading them
as one opaque tag means a single-genre "Pop" track and a multi-genre
track listing Pop both fail to share the Pop key, so the tags axis
in similarity scoring (weight 0.7!) returned 0 in nearly all cases
on real libraries. Result: contextual scoring couldn't differentiate.
Add splitGenres() that splits on ; and , and trims whitespace;
strings without a delimiter come back as a single-element slice
(so the existing single-genre cases keep working unchanged).
Concatenated-without-separator output ("ElectronicComplextroGlitch
Hop") stays opaque — that needs a genre dictionary, out of scope.
2. Play-radio button on TrackRow
playRadio() was only wired to the click handler on /search and
/search/tracks, leaving /library/liked, album pages, and search
results' inner clicks with no way to start a radio. Adds a small
radio button to TrackRow (between LikeButton and the +queue button).
Click → playRadio(track.id), with stopPropagation so the row's own
activate handler doesn't also fire.
Tests:
- 3 new sessionvector tests: TestSplitGenres, multi-genre semicolon
split, no-separator stays opaque. Existing single-genre tests pass
unchanged.
- 1 new TrackRow test: radio button calls playRadio with track id and
does NOT trigger row play.
Verified locally: go -short -race ./... clean, golangci-lint clean,
svelte-check 0/0, 175 vitest tests (was 174), web build succeeds.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The actual cause of the search-field focus loss: SvelteKit's `goto()`
defaults to resetting focus on navigation, so each debounced URL update
blurred the input mid-typing. Pass `keepFocus: true` to both `goto()`
calls in `navigate()`.
The earlier `bind:value` change in 6931358 didn't address this — it was
solving a hypothetical DOM-write-side cursor-jump issue rather than the
real focus-reset behavior. Leaving `bind:value` in place since it's
still cleaner Svelte 5.
Tests updated to match the new goto args.
The previous one-way `{value}` + manual `oninput` rewrite let Svelte
re-set the input's DOM value whenever navigation completed, which moved
the cursor (and on some browsers, blurred the input) mid-typing. Switch
to Svelte 5's `bind:value` — it skips DOM writes when the JS value
already matches the DOM, so the user's typing isn't disturbed.
The URL-resync `\$effect` is unchanged: it still updates `value` for
back/forward navigation but is a no-op during normal typing because the
debounce already wrote the matching URL.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
CI's golangci-lint (`gofmt -s`) on Go 1.23 flagged inline comments after
the SessionVector literals in TestContextualMatchScore_TakesMax because
the variable-length values produced misaligned trailing columns. Local
gofmt on Go 1.26 was lenient. Replace with a leading comment that covers
all three rows; same intent, no alignment dance.
Wire contextual scoring end-to-end: add ContextWeight to RecommendationConfig
(struct + Default()), fetch the user's current session vector in handleRadio
via loadCurrentSessionVector (cold-start returns Seed sentinel), and pass it
as the 6th arg to LoadCandidates so ContextualMatchScore flows into Shuffle.
Code review caught four tests using `got, _` to discard load errors. A DB
failure would have surfaced as a misleading "wrong score" assertion rather
than a clear "load failed" message. Match the t.Fatalf pattern used by the
other LoadCandidates tests in this file.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds a 6th argument (currentVector SessionVector) to LoadCandidates,
bulk-fetches the user's active contextual_likes in one query, and sets
ContextualMatchScore on each candidate's ScoringInputs via max-similarity
over that candidate's like vectors. Adds helper + 6 integration tests
covering no-likes, single match, multi-like max, soft-delete filtering,
seed-like filtering, and seed-current short-circuit. Adds contextual_likes
to the TRUNCATE list in testPool to prevent cross-test leakage.
internal/api/radio.go is intentionally left broken (Task 6 fixes it).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Eight-task TDD-shaped plan covering pure Similarity + ContextualMatchScore
functions, Score extension with ContextWeight=2.0, two new sqlc queries
(ListActiveContextualLikesForUser, GetCurrentSessionVectorForUser),
LoadCandidates signature change to accept currentVector, radio handler
wiring, end-to-end contextual ranking test, and final verification.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Closes M3 — adds the `contextual_match_score` term to the scoring formula
via weighted-Jaccard similarity (tags 0.7, artists 0.3) over the user's
contextual_likes. Reads the current session vector from the most recent
open play_event (populated by sub-plan #2). Cold-start paths collapse to
zero contribution, preserving v1 behavior.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
bvandeusen
