8db59d4bbe
Data layer (Zod-gated, behind StorageRepo; web Dexie v2 + native expo-sqlite user_version migrations, both with no-data-loss segment-id backfill): - Course entity + courseId/lectureDate/instructor/location/tags/lectureNumber on transcripts; listCourses/createCourse/updateCourse/deleteCourse, listByCourse, reassign. Stable per-segment ids assigned on create/update + backfilled. - Persisted source audio (Dexie blob / expo-file-system) via putMedia/getMediaUrl; the editor replays it after reload. UI/stores: - Pre-capture sheet (title + course + lecture date + language) before transcribing. - Course-aware library home (All/Unsorted/per-course filter) + Courses screen. - coursesStore; transcriptsStore course filter + reassign; transcribeStore threads course/date and persists media. 16 web-repo tests incl. v1->v2 migration no-data-loss; 148 total green; web export OK. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
325 lines
12 KiB
TypeScript
325 lines
12 KiB
TypeScript
// Contract tests for the storage layer, run against the web (Dexie) repo.
|
|
//
|
|
// fake-indexeddb/auto polyfills IndexedDB into the Node global so Dexie works
|
|
// under vitest. We test the Dexie repo as the reference implementation of the
|
|
// StorageRepo contract; the native expo-sqlite repo (native-only) is NOT
|
|
// imported here because expo-sqlite cannot load outside a device/simulator.
|
|
//
|
|
// IMPORTANT: repo.web.ts is *not* statically imported. To exercise the v1->v2
|
|
// migration we must first seed a v1-shaped DB with a raw Dexie connection, and
|
|
// only THEN import repo.web (whose module-eval opens the DB at v2, triggering
|
|
// the upgrade). The dynamically-imported repo is stashed in a module-level
|
|
// binding so all subsequent tests reuse it.
|
|
import 'fake-indexeddb/auto';
|
|
|
|
import Dexie from 'dexie';
|
|
import { describe, it, expect, beforeAll, beforeEach } from 'vitest';
|
|
import {
|
|
parseDraft,
|
|
type TranscriptDraft,
|
|
type StorageRepo,
|
|
} from './index';
|
|
|
|
// Populated by the migration beforeAll (which seeds v1 then imports repo.web).
|
|
let repo: StorageRepo;
|
|
|
|
// The v1 record we seed: NO segment ids, NO courseId, but WITH a searchText
|
|
// column (exactly the legacy shape the v1 repo persisted).
|
|
const V1_ID = 't_legacy';
|
|
const v1Row = {
|
|
id: V1_ID,
|
|
title: 'Legacy Lecture',
|
|
createdAt: 1000,
|
|
updatedAt: 1000,
|
|
durationSec: 42.5,
|
|
modelId: 'tiny.en' as const,
|
|
segmentCount: 2,
|
|
segments: [
|
|
{ start: 0, end: 2, text: 'legacy hello' },
|
|
{ start: 2, end: 5, text: 'legacy world', confidence: 0.8 },
|
|
],
|
|
searchText: 'legacy lecture\nlegacy hello\nlegacy world',
|
|
};
|
|
|
|
beforeAll(async () => {
|
|
// 1) Seed a v1 DB with a raw Dexie connection (only the v1 store/schema).
|
|
const v1 = new Dexie('wisp');
|
|
v1.version(1).stores({ transcripts: 'id, createdAt' });
|
|
await v1.open();
|
|
await v1.table('transcripts').put(v1Row);
|
|
v1.close();
|
|
|
|
// 2) Now import repo.web, which opens 'wisp' at v2 and runs the upgrade.
|
|
const mod = await import('./repo.web');
|
|
repo = mod.repo;
|
|
|
|
// Stub object-URL creation so getMediaUrl returns a string under Node.
|
|
globalThis.URL.createObjectURL = () => 'blob:mock';
|
|
});
|
|
|
|
// Helper: a minimal valid draft, overridable per test.
|
|
function makeDraft(over: Partial<TranscriptDraft> = {}): TranscriptDraft {
|
|
return {
|
|
title: 'My Recording',
|
|
durationSec: 12.5,
|
|
modelId: 'tiny.en',
|
|
segments: [
|
|
{ start: 0, end: 2, text: 'hello world' },
|
|
{ start: 2, end: 5, text: 'second segment', confidence: 0.9 },
|
|
],
|
|
...over,
|
|
};
|
|
}
|
|
|
|
describe('v1 -> v2 migration (no data loss)', () => {
|
|
it('keeps the legacy transcript, backfills segment ids, leaves it Unsorted', async () => {
|
|
const fetched = await repo.get(V1_ID);
|
|
expect(fetched).toBeDefined();
|
|
|
|
// Identity-preserving fields survive untouched.
|
|
expect(fetched!.title).toBe('Legacy Lecture');
|
|
expect(fetched!.durationSec).toBe(42.5);
|
|
|
|
// Segment text/timing is identical to the v1 body...
|
|
expect(fetched!.segments.map((s) => s.text)).toEqual([
|
|
'legacy hello',
|
|
'legacy world',
|
|
]);
|
|
expect(fetched!.segments.map((s) => ({ start: s.start, end: s.end }))).toEqual([
|
|
{ start: 0, end: 2 },
|
|
{ start: 2, end: 5 },
|
|
]);
|
|
expect(fetched!.segments[1]!.confidence).toBe(0.8);
|
|
|
|
// ...but every segment now carries a backfilled stable id.
|
|
for (const s of fetched!.segments) {
|
|
expect(s.id).toMatch(/^s_/);
|
|
}
|
|
|
|
// No courseId was set by v1 => the row is "Unsorted" (null/undefined).
|
|
expect(fetched!.courseId == null).toBe(true);
|
|
|
|
// And it shows up under the Unsorted bucket.
|
|
const unsorted = await repo.listByCourse(null);
|
|
expect(unsorted.map((m) => m.id)).toContain(V1_ID);
|
|
});
|
|
});
|
|
|
|
describe('StorageRepo (Dexie web impl)', () => {
|
|
// Wipe transcripts/courses/media between tests so ordering/search/course
|
|
// assertions are deterministic. (Runs after the migration beforeAll.)
|
|
beforeEach(async () => {
|
|
const all = await repo.list();
|
|
await Promise.all(all.map((m) => repo.remove(m.id)));
|
|
const courses = await repo.listCourses();
|
|
await Promise.all(courses.map((c) => repo.deleteCourse(c.id)));
|
|
});
|
|
|
|
it('create -> get round-trips segments and metadata, assigns segment ids', async () => {
|
|
const created = await repo.create(makeDraft());
|
|
|
|
expect(created.id).toMatch(/^t_/);
|
|
expect(created.createdAt).toBeTypeOf('number');
|
|
expect(created.updatedAt).toBe(created.createdAt);
|
|
expect(created.segmentCount).toBe(2);
|
|
// courseId defaults to null ("Unsorted") when absent.
|
|
expect(created.courseId).toBeNull();
|
|
expect(created.hasMedia).toBe(false);
|
|
// Every segment got a stable id assigned on write.
|
|
for (const s of created.segments) {
|
|
expect(s.id).toMatch(/^s_/);
|
|
}
|
|
|
|
const fetched = await repo.get(created.id);
|
|
expect(fetched).toBeDefined();
|
|
expect(fetched!.title).toBe('My Recording');
|
|
expect(fetched!.durationSec).toBe(12.5);
|
|
expect(fetched!.modelId).toBe('tiny.en');
|
|
// Text/timing survive the round-trip exactly.
|
|
expect(fetched!.segments.map((s) => ({ start: s.start, end: s.end, text: s.text }))).toEqual([
|
|
{ start: 0, end: 2, text: 'hello world' },
|
|
{ start: 2, end: 5, text: 'second segment' },
|
|
]);
|
|
expect(fetched!.segments[1]!.confidence).toBe(0.9);
|
|
});
|
|
|
|
it('get returns undefined for an unknown id', async () => {
|
|
expect(await repo.get('t_nope')).toBeUndefined();
|
|
});
|
|
|
|
it('list returns metadatas newest-first (without segments)', async () => {
|
|
const a = await repo.create(makeDraft({ title: 'first' }));
|
|
await new Promise((r) => setTimeout(r, 2));
|
|
const b = await repo.create(makeDraft({ title: 'second' }));
|
|
await new Promise((r) => setTimeout(r, 2));
|
|
const c = await repo.create(makeDraft({ title: 'third' }));
|
|
|
|
const metas = await repo.list();
|
|
expect(metas.map((m) => m.id)).toEqual([c.id, b.id, a.id]);
|
|
expect((metas[0] as unknown as Record<string, unknown>).segments).toBeUndefined();
|
|
expect(metas[0]!.segmentCount).toBe(2);
|
|
});
|
|
|
|
it('update merges title/segments/courseId and re-ids segments + re-derives search', async () => {
|
|
const course = await repo.createCourse({ name: 'Physics' });
|
|
const created = await repo.create(
|
|
makeDraft({ title: 'Old Title', segments: [{ start: 0, end: 1, text: 'apple' }] }),
|
|
);
|
|
const originalUpdatedAt = created.updatedAt;
|
|
await new Promise((r) => setTimeout(r, 2));
|
|
|
|
const updated = await repo.update(created.id, {
|
|
title: 'Brand New Title',
|
|
courseId: course.id,
|
|
segments: [
|
|
{ start: 0, end: 1, text: 'banana' },
|
|
{ start: 1, end: 2, text: 'cherry' },
|
|
],
|
|
});
|
|
|
|
expect(updated.title).toBe('Brand New Title');
|
|
expect(updated.courseId).toBe(course.id);
|
|
expect(updated.segmentCount).toBe(2);
|
|
expect(updated.updatedAt).toBeGreaterThan(originalUpdatedAt);
|
|
// New segments got ids.
|
|
for (const s of updated.segments) {
|
|
expect(s.id).toMatch(/^s_/);
|
|
}
|
|
|
|
// Old text is no longer findable...
|
|
expect(await repo.search('apple')).toHaveLength(0);
|
|
// ...and the new text + new title are.
|
|
expect((await repo.search('cherry')).map((m) => m.id)).toEqual([created.id]);
|
|
expect((await repo.search('brand new')).map((m) => m.id)).toEqual([created.id]);
|
|
});
|
|
|
|
it('update throws for an unknown id', async () => {
|
|
await expect(repo.update('t_missing', { title: 'x' })).rejects.toThrow();
|
|
});
|
|
|
|
it('search matches title and segment text case-insensitively', async () => {
|
|
const t = await repo.create(
|
|
makeDraft({
|
|
title: 'Quarterly MEETING notes',
|
|
segments: [{ start: 0, end: 3, text: 'Discuss the BUDGET forecast' }],
|
|
}),
|
|
);
|
|
|
|
expect((await repo.search('meeting')).map((m) => m.id)).toEqual([t.id]);
|
|
expect((await repo.search('budget')).map((m) => m.id)).toEqual([t.id]);
|
|
expect(await repo.search('zzz-not-present')).toHaveLength(0);
|
|
});
|
|
|
|
it('remove deletes a transcript (and its media)', async () => {
|
|
const created = await repo.create(makeDraft());
|
|
await repo.putMedia(created.id, { data: new ArrayBuffer(4), mime: 'audio/mpeg' });
|
|
expect(await repo.get(created.id)).toBeDefined();
|
|
|
|
await repo.remove(created.id);
|
|
expect(await repo.get(created.id)).toBeUndefined();
|
|
expect(await repo.list()).toHaveLength(0);
|
|
// Media is gone too: no playable URL remains.
|
|
expect(await repo.getMediaUrl(created.id)).toBeUndefined();
|
|
});
|
|
|
|
// --- courses -------------------------------------------------------------
|
|
|
|
it('listByCourse splits Unsorted vs a course', async () => {
|
|
const course = await repo.createCourse({ name: 'Bio 101' });
|
|
const inCourse = await repo.create(makeDraft({ title: 'lecture', courseId: course.id }));
|
|
const unsorted = await repo.create(makeDraft({ title: 'loose note' }));
|
|
|
|
const courseRows = await repo.listByCourse(course.id);
|
|
expect(courseRows.map((m) => m.id)).toEqual([inCourse.id]);
|
|
|
|
const unsortedRows = await repo.listByCourse(null);
|
|
expect(unsortedRows.map((m) => m.id)).toEqual([unsorted.id]);
|
|
});
|
|
|
|
it('createCourse / listCourses (alphabetical) / updateCourse', async () => {
|
|
const zebra = await repo.createCourse({ name: 'Zebra Studies', code: 'ZOO-1' });
|
|
const apple = await repo.createCourse({ name: 'Apple Theory' });
|
|
|
|
expect(apple.id).toMatch(/^c_/);
|
|
expect(apple.createdAt).toBeTypeOf('number');
|
|
|
|
// Alphabetical by name.
|
|
const listed = await repo.listCourses();
|
|
expect(listed.map((c) => c.name)).toEqual(['Apple Theory', 'Zebra Studies']);
|
|
|
|
const renamed = await repo.updateCourse(zebra.id, { name: 'Aardvark Studies' });
|
|
expect(renamed.name).toBe('Aardvark Studies');
|
|
expect(renamed.code).toBe('ZOO-1'); // unspecified fields preserved
|
|
expect(renamed.updatedAt).toBeGreaterThanOrEqual(zebra.updatedAt);
|
|
|
|
const relisted = await repo.listCourses();
|
|
expect(relisted.map((c) => c.name)).toEqual(['Aardvark Studies', 'Apple Theory']);
|
|
});
|
|
|
|
it('deleteCourse reassigns its transcripts to Unsorted, then removes the course', async () => {
|
|
const course = await repo.createCourse({ name: 'Doomed' });
|
|
const t = await repo.create(makeDraft({ title: 'orphan-to-be', courseId: course.id }));
|
|
|
|
await repo.deleteCourse(course.id);
|
|
|
|
// The course is gone.
|
|
expect((await repo.listCourses()).map((c) => c.id)).not.toContain(course.id);
|
|
// The transcript survives, now Unsorted.
|
|
const fetched = await repo.get(t.id);
|
|
expect(fetched).toBeDefined();
|
|
expect(fetched!.courseId).toBeNull();
|
|
expect((await repo.listByCourse(null)).map((m) => m.id)).toContain(t.id);
|
|
});
|
|
|
|
it('reassign moves a transcript between courses and to Unsorted', async () => {
|
|
const course = await repo.createCourse({ name: 'Chemistry' });
|
|
const t = await repo.create(makeDraft());
|
|
expect(t.courseId).toBeNull();
|
|
|
|
await repo.reassign(t.id, course.id);
|
|
expect((await repo.listByCourse(course.id)).map((m) => m.id)).toEqual([t.id]);
|
|
|
|
await repo.reassign(t.id, null);
|
|
expect((await repo.listByCourse(course.id))).toHaveLength(0);
|
|
expect((await repo.listByCourse(null)).map((m) => m.id)).toContain(t.id);
|
|
});
|
|
|
|
// --- media ---------------------------------------------------------------
|
|
|
|
it('putMedia flips hasMedia true and getMediaUrl returns a url; removeMedia flips it false', async () => {
|
|
const t = await repo.create(makeDraft());
|
|
expect(t.hasMedia).toBe(false);
|
|
|
|
await repo.putMedia(t.id, { data: new ArrayBuffer(8), mime: 'audio/mpeg' });
|
|
|
|
const afterPut = await repo.get(t.id);
|
|
expect(afterPut!.hasMedia).toBe(true);
|
|
expect(await repo.getMediaUrl(t.id)).toBe('blob:mock');
|
|
|
|
await repo.removeMedia(t.id);
|
|
const afterRemove = await repo.get(t.id);
|
|
expect(afterRemove!.hasMedia).toBe(false);
|
|
expect(await repo.getMediaUrl(t.id)).toBeUndefined();
|
|
});
|
|
|
|
// --- validation gatekeeping ---------------------------------------------
|
|
|
|
it('parseDraft rejects a NaN/Infinity duration', () => {
|
|
expect(() => parseDraft(makeDraft({ durationSec: NaN }))).toThrow();
|
|
expect(() => parseDraft(makeDraft({ durationSec: Infinity }))).toThrow();
|
|
});
|
|
|
|
it('parseDraft rejects a non-finite segment time', () => {
|
|
expect(() =>
|
|
parseDraft(makeDraft({ segments: [{ start: 0, end: Infinity, text: 'x' }] })),
|
|
).toThrow();
|
|
expect(() =>
|
|
parseDraft(makeDraft({ segments: [{ start: NaN, end: 1, text: 'x' }] })),
|
|
).toThrow();
|
|
});
|
|
|
|
it('create rejects an invalid draft (defense in depth)', async () => {
|
|
await expect(repo.create(makeDraft({ durationSec: -1 }))).rejects.toThrow();
|
|
});
|
|
});
|