soulsync/webui
Broque Thomas 258905ff5c Fix: duplicate tracks in albums with Japanese / CJK titles (#722)
Reporter @Sokhii: downloading the Mushoku Tensei Original
Soundtrack II via Apple Music metadata + Tidal download
produced duplicate library entries — same audio file landed
under multiple track positions in the album view.

Root cause (verified by direct probe + isolated repro):
``MusicMatchingEngine.normalize_string`` correctly skipped
unidecode for CJK text (kanji→pinyin would have produced
gibberish — see the inline comment at line 74-76), but then
ran ``re.sub(r'[^a-z0-9\s$]', '', text)`` which stripped EVERY
CJK character. Every Japanese title normalised to ``''``.
``similarity_score`` has an early-out guard
    if not str1 or not str2: return 0.0
so EVERY CJK-vs-CJK title comparison returned 0.000.

Downstream effect: the matcher fell back to duration+artist
alone. For an OST album with 24 tracks all by the same artist
with similar durations, multiple iTunes track queries landed
on the SAME Tidal candidate. SoulSync wrote each download to
a different output filename (per the iTunes track position),
so on disk there were N copies of the same audio under
different track numbers. The user's library showed 34 entries
for an album with 24 actual tracks.

Probed iTunes album 1753240110 directly — 24 distinct tracks,
zero (disc, track_number) collisions, both US + JP storefronts.
So the duplicate origin was definitely downstream of metadata
fetch.

Fix: when CJK is detected upstream, the alphanumeric-strip step
also preserves CJK Unified Ideographs + radicals
(⺀-鿿), Hiragana + Katakana (぀-ヿ), Halfwidth
/ Fullwidth forms (＀-￯), and Hangul syllables
(가-힯). CJK titles now produce a comparable normalised
form instead of an empty string. ``similarity_score`` works as
intended:

  '命の灯火' vs '命の灯火' → 1.000  (was 0.000)
  '命の灯火' vs '無職転生' → 0.000  (was 0.000, but now from
                                       actual char comparison
                                       not from the empty-string
                                       guard)

Latin-only normalisation is completely unchanged. ``has_cjk``
is False for Latin input, so both the CJK-lowercase branch AND
the new CJK-preserve strip branch are skipped — Latin titles
go through the original unidecode + lowercase + strip path
verbatim. Tested via 4 regression tests that pin the Latin
baseline (simple, unidecode target, $-preservation, identical
+ different similarity scores).

16 new unit tests in ``tests/test_matching_engine_cjk.py``:
- Kanji / Hiragana / Katakana / Hangul / Chinese all survive
- CJK-only strip still removes Latin punctuation in the
  CJK branch
- Mixed Latin + CJK lowercases the Latin half
- Identical CJK titles → 1.0
- Disjoint CJK titles → near 0
- Partially overlapping CJK titles → midrange
- CJK doesn't falsely match unrelated Latin
- 4 Latin-baseline regression pins
- Real-world Mushoku Tensei OST scenario

371 text + imports + new CJK tests pass after the fix.
2026-05-28 08:53:43 -07:00
..
docs/migration docs(import): align migration docs 2026-05-24 21:17:22 +03:00
src Fix Zustand shallow import 2026-05-24 14:47:29 -07:00
static Fix: duplicate tracks in albums with Japanese / CJK titles (#722) 2026-05-28 08:53:43 -07:00
tests fix(webui): preserve import tab refresh URLs 2026-05-24 21:17:20 +03:00
.gitignore Initial Vite app scaffolding & issues page impl 2026-05-13 22:24:46 +03:00
.oxfmtrc.json Split webui tooling into separate configs 2026-05-13 22:26:25 +03:00
.oxlintrc.json Split webui tooling into separate configs 2026-05-13 22:26:25 +03:00
index.html Sync page: collapse tabs to brand-logo chips with active label pill 2026-05-27 22:04:28 -07:00
package-lock.json feat(webui): migrate import route to React 2026-05-24 21:11:40 +03:00
package.json feat(webui): migrate import route to React 2026-05-24 21:11:40 +03:00
playwright.config.ts Initial Vite app scaffolding & issues page impl 2026-05-13 22:24:46 +03:00
README.md docs(webui): group migration planning docs 2026-05-23 21:22:44 +03:00
tsconfig.json Split webui tooling into separate configs 2026-05-13 22:26:25 +03:00
vite.config.ts Split webui tooling into separate configs 2026-05-13 22:26:25 +03:00
vitest.config.ts Split webui tooling into separate configs 2026-05-13 22:26:25 +03:00
vitest.setup.ts feat(webui): expose shell status in root context 2026-05-23 21:23:32 +03:00

WebUI Hybrid Rendering

SoulSync's web UI is in a transition phase:

  • most pages still render through the legacy vanilla JS shell
  • /issues is rendered by the new React app
  • a small shell bridge keeps both runtimes aware of the active page, profile context, and navigation state

How It Fits Together

flowchart LR
    Browser["Browser parses /webui/index.html"]
    Legacy["Legacy shell scripts\n(core.js -> ... -> init.js)"]
    Bridge["shell-bridge.js\nwindow.SoulSyncWebShellBridge"]
    React["Vite React app\nsrc/app/main.tsx"]
    Router["TanStack Router\nwindow.SoulSyncWebRouter"]

    Browser --> Legacy
    Browser --> React
    Legacy --> Bridge
    React --> Router
    Router --> Bridge
    Bridge --> Legacy

Runtime Roles

  • webui/static/init.js

    • boots the legacy shell
    • selects the active profile
    • handles the legacy page loading flow
  • webui/static/shell-bridge.js

    • owns the browser-side bridge object
    • exposes window.SoulSyncWebShellBridge
    • owns the shared page chrome and route handoff helpers
  • webui/src/app/main.tsx

    • mounts the React app
    • binds window.SoulSyncWebRouter
  • webui/src/platform/shell/route-controllers.tsx

    • listens for bridge readiness
    • keeps React pages aligned with the shell

Load Order

The current order in index.html matters:

  1. legacy shell scripts load first
  2. init.js sets up the shell runtime
  3. shell-bridge.js publishes the bridge and shared chrome helpers after the shell state exists
  4. the Vite React app is injected through {{ vite_assets('body') }} and boots as a module after parsing

That order avoids load-time references to missing globals and keeps the React side able to react to bridge readiness events. The React entry can start fetching early, but the shell bridge and legacy globals are already available by the time the React runtime starts acting on them.

Notes

  • The bridge is intentionally small and browser-only.
  • This is the start of the migration, not a full replacement of the legacy shell.
  • When adding another React page, check whether it needs:
    • a route entry in webui/src/platform/shell/route-manifest.ts
    • bridge typings in webui/src/platform/shell/globals.d.ts
    • a legacy fallback path in webui/static/init.js
    • bridge glue or handoff logic in webui/static/shell-bridge.js

Folder Layout

The React webui uses a small set of predictable folders so route slices stay easy to extend, test, and understand.

webui/src/
  app/         React bootstrap, router, query client, shared API client
  components/  Shared UI primitives
  platform/    Shell bridge and browser/platform integration
  routes/      Route-local code and TanStack Router pages
  test/        Shared test utilities and setup helpers

Migration planning docs live under webui/docs/migration/.

  • keep the high-level route backlog there
  • add one route-specific sketch per migration task
  • keep migration notes close to the WebUI code rather than the repo root

Route Slices

  • Keep route-specific code inside webui/src/routes/<route>/.
  • Put the routing entry in route.tsx.
  • Put route-local UI in a -ui/ folder.
  • Prefix non-routing files with - so TanStack Router ignores them.
  • Keep the route slice small and cohesive.
  • Prefer a few files with clear responsibilities over many tiny files with overlapping names.

Example:

webui/src/routes/issues/
  route.tsx
  -issues.types.ts
  -issues.api.ts
  -issues.helpers.ts
  -issues.api.test.ts
  -issues.helpers.test.ts
  -ui/
    issues-page.tsx
    issue-detail-modal.tsx
    issue-domain-host.tsx

The initial issues slice is the model to follow:

  • -issues.api.ts holds request code and query options
  • -issues.helpers.ts holds pure normalization and formatting
  • -issues.types.ts holds shared types
  • -ui/ holds the page, modal, and legacy handoff UI

Shared Code

  • Put reusable UI in webui/src/components/.
  • Put shell integration in webui/src/platform/.
  • Put bootstrap and app-wide wiring in webui/src/app/.
  • Move code up a level only when it is genuinely shared.
  • Avoid creating new conventions that overlap with existing ones.

Testing Choices

We have a lot of testing tools available, but we do not need all of them for every feature.

  • Use plain unit tests for pure functions and small transforms.
  • Use React component or route tests when the behavior lives in the UI or router.
  • Use MSW-backed tests when request shape, response handling, or error handling matters.
  • Use Playwright when the behavior is best proven end-to-end with the server and browser together.
  • Prefer the smallest test setup that still proves the thing that can regress.

Development

The repo root now owns the full local-dev instructions. Start there for the portable launcher and backend/frontend setup:

  1. README.md for the end-to-end dev flow
  2. npm run check and npm run fix for React-side linting and formatting