Fix #32: Architecture: Deepen the result handoff: one module at the route seam

[Kingson4Wu]

Closes #32

Generated by RalphWorks. Execution and artifacts. Review the patch and required checks before merging.

[Kingson4Wu]

Please revise this PR before merge: (1) clear() in useResultStore has no production caller, so remove it and its test; acceptance criterion 5 explicitly requires unused clearResult to be called or removed. Keep the public interface to submit and load. (2) Revert the unrelated Vitest upgrade and package-lock churn; retain the versions from main. Then rerun npm test -- --run && npm run build.

[Kingson4Wu]

PR Review: Consolidate result storage into useResultStore

Scope: Replaces useHistory.js (localStorage only) and absorbs formatMbtiType.js into a single useResultStore.js that owns the sessionStorage handoff (mbti_result), the localStorage history (mbti_lab_result), type normalization, entry validation, and fresh-then-history precedence. Views are updated; formatMbtiType.test.js is replaced by useResultStore.test.js; one new ResultView regression test is added.

Verification performed

• npx vitest run → 7 files / 78 tests passed
• npm run build → succeeded (pre-existing >500 kB chunk warning only)
• grep across src/ and tests/ → no dangling imports of useHistory or formatMbtiType; no raw reads of mbti_result / mbti_lab_result outside the new module
• Backward compatibility: the saved result shape ({ type, percents } from useTest.js) has been stable since the initial commit (dd54182), and HISTORY_KEY keeps the old mbti_lab_result name, so existing users' persisted results pass the new validation and still load.
• Refresh behavior on /result is preserved: submit() writes both storages, and after load() consumes the session entry, refresh falls back to an identical history entry.

Findings

Overall this is a clean, well-tested refactor that fixes the reported crash (invalid session entry reaching ResultCard). Findings below are minor / non-blocking.

1. Dead code: clear() is never called — src/composables/useResultStore.js:89-92
   No view or test-other-than-its-own invokes clear(). The old clearResult was equally unused, so this is pre-existing API surface carried forward. Either wire it to a "clear history" affordance or drop it.

2. submit() performs no validation, load() validates strictly — src/composables/useResultStore.js:73-76 vs :36-67
   If a caller ever submits a malformed result, it is persisted to history and then silently rejected on every subsequent load() — the "last result" feature stays broken until a valid test overwrites it. Current callers (useTest → TestView.vue:52) always produce valid shapes, so this is not a live bug, but validating in submit (or returning the parsed entry from it) would fail fast instead.

3. Minor test-coverage regression: whitespace trimming — deleted tests/formatMbtiType.test.js (cases ' enfp ', '\tistj\n') vs tests/useResultStore.test.js:154-158
normalizeType still trims, but only the lowercase case ('infp') is re-covered; the trim cases were dropped and normalizeType is not exported, so they can only be exercised through load(). A one-line putSession({ type: ' enfp ', ... }) case would restore parity.

4. Note (intentional, no change requested): HomeView now consumes the fresh handoff — src/views/HomeView.vue:62
load() is read-and-consume, so a Home visit eats any not-yet-consumed session entry. In current flows both storages hold identical entries after submit, so this is benign and is exactly the mechanism that prevents a stale session entry from shadowing newer history (documented in the module header and covered by the "tab 2" test at tests/useResultStore.test.js:100-118). Flagging only so reviewers are aware of the semantic change to HomeView.

Conclusion

Approve. The core fix is correct and regression-tested (tests/ResultView.test.js:100-110 proves a corrupted session entry renders the empty state instead of crashing ResultCard), storage keys are backward compatible, and all 78 tests plus the production build pass. Findings 1-3 are optional polish.

RalphWorks review run: https://github.com/OctopusGarage/mbti-lab/actions/runs/35680692179

[Kingson4Wu]

RalphWorks addressed PR feedback in f9ba9f0. Run: https://github.com/OctopusGarage/mbti-lab/actions/runs/35680753178