0d3cb498a3
Deploy local.promptfoo.app / Deploy to Cloudflare Pages (push) Waiting to run
Test and Publish Multi-arch Docker Image / test (push) Waiting to run
Test and Publish Multi-arch Docker Image / build-docker-and-push-digests (map[digest-suffix:linux-amd64 platform:linux/amd64 runner:ubuntu-latest]) (push) Blocked by required conditions
Test and Publish Multi-arch Docker Image / build-docker-and-push-digests (map[digest-suffix:linux-arm64 platform:linux/arm64 runner:ubuntu-24.04-arm]) (push) Blocked by required conditions
Test and Publish Multi-arch Docker Image / merge-docker-digests (push) Blocked by required conditions
Test and Publish Multi-arch Docker Image / Attest Multi-arch Image (push) Blocked by required conditions
Validate Renovate Config / Validate Renovate Configuration (push) Waiting to run
CI / Shell Format Check (push) Has been cancelled
CI / Check Ruby (3.4) (push) Has been cancelled
CI / CI Config (push) Has been cancelled
CI / Test on Node ${{ matrix.node }} and ${{ matrix.os }}${{ matrix.shard && format(' (shard {0}/3)', matrix.shard) || '' }} (push) Has been cancelled
CI / Build on Node ${{ matrix.node }} (push) Has been cancelled
CI / Style Check (push) Has been cancelled
CI / Generate Assets (push) Has been cancelled
CI / Check Python (3.14) (push) Has been cancelled
CI / Check Python (3.9) (push) Has been cancelled
CI / Build Docs (push) Has been cancelled
CI / Code Scan Action (push) Has been cancelled
CI / Site tests (push) Has been cancelled
CI / webui tests (push) Has been cancelled
CI / Run Integration Tests (push) Has been cancelled
CI / Run Smoke Tests (push) Has been cancelled
CI / Go Tests (push) Has been cancelled
CI / Share Test (push) Has been cancelled
CI / Redteam (Production API) (push) Has been cancelled
CI / Redteam (Staging API) (push) Has been cancelled
CI / GitHub Actions Lint (push) Has been cancelled
CI / Check Ruby (3.0) (push) Has been cancelled
release-please / release-please (push) Has been cancelled
release-please / build (push) Has been cancelled
release-please / publish-npm (push) Has been cancelled
release-please / publish-npm-backfill (push) Has been cancelled
release-please / docker (push) Has been cancelled
release-please / publish-code-scan-action (push) Has been cancelled
release-please / attest-code-scan-action (push) Has been cancelled
136 lines
4.7 KiB
TypeScript
136 lines
4.7 KiB
TypeScript
import { afterEach, beforeAll, beforeEach, describe, expect, it, vi } from 'vitest';
|
|
import { getDb } from '../../src/database/index';
|
|
import { updateSignalFileForDeletedEvals } from '../../src/database/signal';
|
|
import { spansTable, tracesTable } from '../../src/database/tables';
|
|
import { runDbMigrations } from '../../src/migrate';
|
|
import Eval from '../../src/models/eval';
|
|
import { TraceStore } from '../../src/tracing/store';
|
|
import { deleteAllEvals, deleteEval, deleteEvals } from '../../src/util/database';
|
|
import EvalFactory from '../factories/evalFactory';
|
|
|
|
vi.mock('../../src/database/signal', async () => {
|
|
const actual = await vi.importActual('../../src/database/signal');
|
|
return {
|
|
...actual,
|
|
updateSignalFile: vi.fn(),
|
|
updateSignalFileForDeletedEvals: vi.fn(),
|
|
};
|
|
});
|
|
|
|
describe('database eval deletion', () => {
|
|
beforeAll(async () => {
|
|
await runDbMigrations();
|
|
});
|
|
|
|
afterEach(() => {
|
|
vi.resetAllMocks();
|
|
});
|
|
|
|
beforeEach(async () => {
|
|
const db = await getDb();
|
|
await db.run('DELETE FROM spans');
|
|
await db.run('DELETE FROM traces');
|
|
await db.run('DELETE FROM eval_results');
|
|
await db.run('DELETE FROM evals_to_datasets');
|
|
await db.run('DELETE FROM evals_to_prompts');
|
|
await db.run('DELETE FROM evals_to_tags');
|
|
await db.run('DELETE FROM evals');
|
|
});
|
|
|
|
async function addTrace(evalId: string, traceId: string) {
|
|
const traceStore = new TraceStore();
|
|
await traceStore.createTrace({
|
|
traceId,
|
|
evaluationId: evalId,
|
|
testCaseId: 'test-case-id',
|
|
});
|
|
await traceStore.addSpans(traceId, [
|
|
{
|
|
spanId: `${traceId}-span`,
|
|
name: 'test-span',
|
|
startTime: 1,
|
|
},
|
|
]);
|
|
}
|
|
|
|
it('deletes traces and spans for a single eval', async () => {
|
|
const eval_ = await EvalFactory.create();
|
|
await addTrace(eval_.id, 'trace-single');
|
|
|
|
await deleteEval(eval_.id);
|
|
|
|
const db = await getDb();
|
|
expect(await Eval.findById(eval_.id)).toBeUndefined();
|
|
expect(await db.select().from(tracesTable).all()).toHaveLength(0);
|
|
expect(await db.select().from(spansTable).all()).toHaveLength(0);
|
|
expect(updateSignalFileForDeletedEvals).toHaveBeenCalledWith([eval_.id]);
|
|
});
|
|
|
|
it('deletes only traces and spans for selected evals', async () => {
|
|
const eval1 = await EvalFactory.create();
|
|
const eval2 = await EvalFactory.create();
|
|
const eval3 = await EvalFactory.create();
|
|
await addTrace(eval1.id, 'trace-bulk-1');
|
|
await addTrace(eval2.id, 'trace-bulk-2');
|
|
await addTrace(eval3.id, 'trace-retained');
|
|
|
|
await deleteEvals([eval1.id, eval2.id]);
|
|
|
|
const db = await getDb();
|
|
expect(await Eval.findById(eval1.id)).toBeUndefined();
|
|
expect(await Eval.findById(eval2.id)).toBeUndefined();
|
|
expect(await Eval.findById(eval3.id)).toBeDefined();
|
|
expect(await db.select({ traceId: tracesTable.traceId }).from(tracesTable).all()).toEqual([
|
|
{ traceId: 'trace-retained' },
|
|
]);
|
|
expect(await db.select({ traceId: spansTable.traceId }).from(spansTable).all()).toEqual([
|
|
{ traceId: 'trace-retained' },
|
|
]);
|
|
expect(updateSignalFileForDeletedEvals).toHaveBeenCalledWith([eval1.id, eval2.id]);
|
|
});
|
|
|
|
it('does not emit a delete signal when called with an empty id list', async () => {
|
|
// An empty deletedEvalIds list is indistinguishable from "all evals deleted" on the
|
|
// client, so deleting zero evals must be a no-op rather than a spurious clear.
|
|
await deleteEvals([]);
|
|
|
|
expect(updateSignalFileForDeletedEvals).not.toHaveBeenCalled();
|
|
});
|
|
|
|
it('deletes traces and spans when deleting all evals', async () => {
|
|
const eval1 = await EvalFactory.create();
|
|
const eval2 = await EvalFactory.create();
|
|
await addTrace(eval1.id, 'trace-all-1');
|
|
await addTrace(eval2.id, 'trace-all-2');
|
|
|
|
await deleteAllEvals();
|
|
|
|
const db = await getDb();
|
|
expect(await Eval.getMany()).toHaveLength(0);
|
|
expect(await db.select().from(tracesTable).all()).toHaveLength(0);
|
|
expect(await db.select().from(spansTable).all()).toHaveLength(0);
|
|
expect(updateSignalFileForDeletedEvals).toHaveBeenCalledWith(undefined);
|
|
});
|
|
|
|
it('handles evals with no traces without error', async () => {
|
|
const eval_ = await EvalFactory.create();
|
|
|
|
await deleteEval(eval_.id);
|
|
|
|
expect(await Eval.findById(eval_.id)).toBeUndefined();
|
|
});
|
|
|
|
it('deletes every span when a trace has multiple spans', async () => {
|
|
const eval_ = await EvalFactory.create();
|
|
await addTrace(eval_.id, 'trace-multi-span');
|
|
await new TraceStore().addSpans('trace-multi-span', [
|
|
{ spanId: 'extra-span', name: 'extra', startTime: 2 },
|
|
]);
|
|
|
|
await deleteEval(eval_.id);
|
|
|
|
const db = await getDb();
|
|
expect(await db.select().from(spansTable).all()).toHaveLength(0);
|
|
});
|
|
});
|