Files
Neon-Desk/lib/server/repos/tasks.test.ts
francy51 ed4420b8db Add atomic task deduplication with partial unique index
- Add partial unique index for active resource-scoped tasks
- Implement createTaskRunRecordAtomic for race-free task creation
- Update findOrEnqueueTask to use atomic insert first
- Add tests for concurrent task creation deduplication
2026-03-15 14:40:38 -04:00

329 lines
10 KiB
TypeScript

import {
afterAll,
beforeAll,
beforeEach,
describe,
expect,
it
} from 'bun:test';
import { mkdtempSync, readFileSync, rmSync } from 'node:fs';
import { tmpdir } from 'node:os';
import { join } from 'node:path';
import { Database } from 'bun:sqlite';
const TEST_USER_ID = 'task-test-user';
let tempDir: string | null = null;
let sqliteClient: Database | null = null;
let tasksRepo: typeof import('./tasks') | null = null;
function resetDbSingletons() {
const globalState = globalThis as typeof globalThis & {
__fiscalSqliteClient?: { close?: () => void };
__fiscalDrizzleDb?: unknown;
};
globalState.__fiscalSqliteClient?.close?.();
globalState.__fiscalSqliteClient = undefined;
globalState.__fiscalDrizzleDb = undefined;
}
function applyMigration(client: Database, fileName: string) {
const sql = readFileSync(join(process.cwd(), 'drizzle', fileName), 'utf8');
client.exec(sql);
}
function ensureUser(client: Database) {
const now = Date.now();
client.exec(`
INSERT OR REPLACE INTO user (id, name, email, emailVerified, image, createdAt, updatedAt, role, banned, banReason, banExpires)
VALUES ('${TEST_USER_ID}', 'Task Test User', 'tasks@example.com', 1, NULL, ${now}, ${now}, NULL, 0, NULL, NULL);
`);
}
function clearTasks(client: Database) {
client.exec('DELETE FROM task_stage_event;');
client.exec('DELETE FROM task_run;');
}
describe('task repos', () => {
beforeAll(async () => {
tempDir = mkdtempSync(join(tmpdir(), 'fiscal-task-repo-'));
const env = process.env as Record<string, string | undefined>;
env.DATABASE_URL = `file:${join(tempDir, 'repo.sqlite')}`;
env.NODE_ENV = 'test';
resetDbSingletons();
sqliteClient = new Database(join(tempDir, 'repo.sqlite'), { create: true });
sqliteClient.exec('PRAGMA foreign_keys = ON;');
for (const file of [
'0000_cold_silver_centurion.sql',
'0001_glossy_statement_snapshots.sql',
'0002_workflow_task_projection_metadata.sql',
'0003_task_stage_event_timeline.sql',
'0004_watchlist_company_taxonomy.sql',
'0005_financial_taxonomy_v3.sql',
'0006_coverage_journal_tracking.sql',
'0007_company_financial_bundles.sql',
'0008_research_workspace.sql',
'0009_task_notification_context.sql',
'0012_company_overview_cache.sql',
'0013_task_active_resource_unique.sql'
]) {
applyMigration(sqliteClient, file);
}
ensureUser(sqliteClient);
tasksRepo = await import('./tasks');
});
afterAll(() => {
sqliteClient?.close();
resetDbSingletons();
if (tempDir) {
rmSync(tempDir, { recursive: true, force: true });
}
});
beforeEach(() => {
if (!sqliteClient) {
throw new Error('sqlite client not initialized');
}
clearTasks(sqliteClient);
});
it('updates same-stage progress without duplicating stage events', async () => {
if (!tasksRepo) {
throw new Error('tasks repo not initialized');
}
const task = await tasksRepo.createTaskRunRecord({
id: 'task-progress',
user_id: TEST_USER_ID,
task_type: 'index_search',
payload: { ticker: 'AAPL' },
priority: 50,
max_attempts: 3,
resource_key: 'index_search:ticker:AAPL'
});
await tasksRepo.markTaskRunning(task.id);
await tasksRepo.updateTaskStage(task.id, 'search.embed', 'Embedding 1 of 3 sources', {
progress: { current: 1, total: 3, unit: 'sources' },
counters: { chunksEmbedded: 12 },
subject: { ticker: 'AAPL', label: 'doc-1' }
});
await tasksRepo.updateTaskStage(task.id, 'search.embed', 'Embedding 2 of 3 sources', {
progress: { current: 2, total: 3, unit: 'sources' },
counters: { chunksEmbedded: 24 },
subject: { ticker: 'AAPL', label: 'doc-2' }
});
const current = await tasksRepo.getTaskByIdForUser(task.id, TEST_USER_ID);
const events = await tasksRepo.listTaskStageEventsForTask(task.id, TEST_USER_ID);
expect(current?.stage_detail).toBe('Embedding 2 of 3 sources');
expect(current?.stage_context?.progress?.current).toBe(2);
expect(events.filter((event) => event.stage === 'search.embed')).toHaveLength(1);
});
it('lists recent tasks by updated_at descending', async () => {
if (!tasksRepo) {
throw new Error('tasks repo not initialized');
}
const first = await tasksRepo.createTaskRunRecord({
id: 'task-first',
user_id: TEST_USER_ID,
task_type: 'refresh_prices',
payload: {},
priority: 50,
max_attempts: 3,
resource_key: 'refresh_prices:portfolio'
});
await Bun.sleep(5);
const second = await tasksRepo.createTaskRunRecord({
id: 'task-second',
user_id: TEST_USER_ID,
task_type: 'portfolio_insights',
payload: {},
priority: 50,
max_attempts: 3,
resource_key: 'portfolio_insights:portfolio'
});
await Bun.sleep(5);
await tasksRepo.updateTaskStage(first.id, 'refresh.fetch_quotes', 'Fetching quotes', {
progress: { current: 1, total: 3, unit: 'tickers' }
});
const tasks = await tasksRepo.listRecentTasksForUser(TEST_USER_ID, 10);
expect(tasks[0]?.id).toBe(first.id);
expect(tasks[1]?.id).toBe(second.id);
});
it('preserves completion and failure detail/context on terminal tasks', async () => {
if (!tasksRepo) {
throw new Error('tasks repo not initialized');
}
const completedTask = await tasksRepo.createTaskRunRecord({
id: 'task-completed',
user_id: TEST_USER_ID,
task_type: 'analyze_filing',
payload: { accessionNumber: '0000320193-26-000001' },
priority: 50,
max_attempts: 3,
resource_key: 'analyze_filing:0000320193-26-000001'
});
await tasksRepo.completeTask(completedTask.id, {
ticker: 'AAPL',
accessionNumber: '0000320193-26-000001',
filingType: '10-Q'
}, {
detail: 'Analysis report generated for AAPL 10-Q 0000320193-26-000001.',
context: {
subject: {
ticker: 'AAPL',
accessionNumber: '0000320193-26-000001',
label: '10-Q'
}
}
});
const failedTask = await tasksRepo.createTaskRunRecord({
id: 'task-failed',
user_id: TEST_USER_ID,
task_type: 'index_search',
payload: { ticker: 'AAPL' },
priority: 50,
max_attempts: 3,
resource_key: 'index_search:ticker:AAPL'
});
await tasksRepo.markTaskFailure(failedTask.id, 'Search indexing could not generate embeddings for AAPL · doc-2. The AI provider returned an empty response.', 'search.embed', {
detail: 'Embedding generation failed.',
context: {
progress: { current: 2, total: 5, unit: 'sources' },
counters: { chunksEmbedded: 20 },
subject: { ticker: 'AAPL', label: 'doc-2' }
}
});
const completed = await tasksRepo.getTaskByIdForUser(completedTask.id, TEST_USER_ID);
const failed = await tasksRepo.getTaskByIdForUser(failedTask.id, TEST_USER_ID);
expect(completed?.stage_detail).toContain('Analysis report generated');
expect(completed?.stage_context?.subject?.ticker).toBe('AAPL');
expect(failed?.stage).toBe('search.embed');
expect(failed?.stage_detail).toBe('Embedding generation failed.');
expect(failed?.error).toContain('Search indexing could not generate embeddings');
expect(failed?.stage_context?.progress?.current).toBe(2);
});
it('atomically deduplicates concurrent task creation for same resource', async () => {
if (!tasksRepo) {
throw new Error('tasks repo not initialized');
}
const resourceKey = 'sync_filings:AAPL';
const concurrentCount = 10;
const results = await Promise.all(
Array.from({ length: concurrentCount }, (_, i) =>
tasksRepo!.createTaskRunRecordAtomic({
id: `task-race-${i}`,
user_id: TEST_USER_ID,
task_type: 'sync_filings',
payload: { ticker: 'AAPL' },
priority: 50,
max_attempts: 3,
resource_key: resourceKey
})
)
);
const createdResults = results.filter((r) => r.created);
const conflictResults = results.filter((r) => !r.created);
expect(createdResults.length).toBe(1);
expect(conflictResults.length).toBe(concurrentCount - 1);
expect(createdResults[0]?.task.resource_key).toBe(resourceKey);
expect(createdResults[0]?.task.status).toBe('queued');
});
it('allows creating new task for same resource after previous task completes', async () => {
if (!tasksRepo) {
throw new Error('tasks repo not initialized');
}
const resourceKey = 'sync_filings:MSFT';
const first = await tasksRepo.createTaskRunRecordAtomic({
id: 'task-first-msft',
user_id: TEST_USER_ID,
task_type: 'sync_filings',
payload: { ticker: 'MSFT' },
priority: 50,
max_attempts: 3,
resource_key: resourceKey
});
expect(first.created).toBe(true);
if (!first.created) throw new Error('Expected task to be created');
await tasksRepo.completeTask(first.task.id, { ticker: 'MSFT' });
const second = await tasksRepo.createTaskRunRecordAtomic({
id: 'task-second-msft',
user_id: TEST_USER_ID,
task_type: 'sync_filings',
payload: { ticker: 'MSFT' },
priority: 50,
max_attempts: 3,
resource_key: resourceKey
});
expect(second.created).toBe(true);
if (!second.created) throw new Error('Expected second task to be created');
expect(second.task.id).not.toBe(first.task.id);
});
it('allows creating tasks without resource key without deduplication', async () => {
if (!tasksRepo) {
throw new Error('tasks repo not initialized');
}
const results = await Promise.all([
tasksRepo.createTaskRunRecordAtomic({
id: 'task-nokey-1',
user_id: TEST_USER_ID,
task_type: 'sync_filings',
payload: { ticker: 'GOOGL' },
priority: 50,
max_attempts: 3,
resource_key: null
}),
tasksRepo.createTaskRunRecordAtomic({
id: 'task-nokey-2',
user_id: TEST_USER_ID,
task_type: 'sync_filings',
payload: { ticker: 'GOOGL' },
priority: 50,
max_attempts: 3,
resource_key: null
})
]);
expect(results[0]?.created).toBe(true);
expect(results[1]?.created).toBe(true);
if (!results[0]?.created || !results[1]?.created) {
throw new Error('Expected both tasks to be created');
}
expect(results[0].task.id).not.toBe(results[1].task.id);
});
});