Files
wehub-resource-sync 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
chore: import upstream snapshot with attribution
2026-07-13 13:24:08 +08:00

372 lines
11 KiB
TypeScript

import { afterEach, beforeEach, describe, expect, it, type Mock, vi } from 'vitest';
import { evaluate } from '../../src/evaluator';
import { nodeEvaluatorRuntime } from '../../src/node/evaluatorRuntime';
import * as evaluatorTracing from '../../src/tracing/evaluatorTracing';
import { getTraceStore } from '../../src/tracing/store';
import { createMockProvider } from '../factories/provider';
import type { EvaluatorRuntime } from '../../src/evaluator/runtime';
import type Eval from '../../src/models/eval';
import type EvalResult from '../../src/models/evalResult';
import type { EvaluateOptions, TestSuite } from '../../src/types/index';
// Mock dependencies
vi.mock('../../src/tracing/store');
const mockFlushOtel = vi.hoisted(() => vi.fn().mockResolvedValue(undefined));
const mockInitializeOtel = vi.hoisted(() => vi.fn());
const mockShutdownOtel = vi.hoisted(() => vi.fn().mockResolvedValue(undefined));
vi.mock('../../src/tracing/otelSdk', () => ({
flushOtel: mockFlushOtel,
initializeOtel: mockInitializeOtel,
shutdownOtel: mockShutdownOtel,
}));
vi.mock('../../src/tracing/otlpReceiver', () => ({
startOTLPReceiver: vi.fn(),
stopOTLPReceiver: vi.fn(),
}));
// Mock evaluatorTracing module
vi.mock('../../src/tracing/evaluatorTracing', () => ({
generateTraceId: vi.fn(() => 'abcdef1234567890abcdef1234567890'),
generateSpanId: vi.fn(() => '0123456789abcdef'),
generateTraceparent: vi.fn((traceId, spanId) => `00-${traceId}-${spanId}-01`),
generateTraceContextIfNeeded: vi.fn(),
startOtlpReceiverIfNeeded: vi.fn(),
stopOtlpReceiverIfNeeded: vi.fn(),
isOtlpReceiverStarted: vi.fn(() => false),
isTracingEnabled: vi.fn((test) => test.metadata?.tracingEnabled === true),
}));
describe('evaluator trace integration', () => {
const mockTraceStore = {
createTrace: vi.fn(),
getTrace: vi.fn(),
};
const mockEval = {
id: 'test-eval-id',
addResult: vi.fn(),
addPrompts: vi.fn(),
fetchResultsByTestIdx: vi.fn(),
setVars: vi.fn(),
setDurationMs: vi.fn(),
results: [],
prompts: [],
persisted: false,
config: {
outputPath: undefined,
},
} as unknown as Eval;
beforeEach(() => {
vi.resetAllMocks();
(getTraceStore as Mock).mockReturnValue(mockTraceStore);
});
afterEach(() => {
vi.restoreAllMocks();
});
it('should pass traceId through to assertions when tracing is enabled', async () => {
// Mock trace creation and retrieval
const testTraceId = 'abcdef1234567890abcdef1234567890';
mockTraceStore.createTrace.mockResolvedValue(undefined);
mockTraceStore.getTrace.mockResolvedValue({
traceId: testTraceId,
spans: [
{
spanId: 'test-span',
name: 'test.operation',
startTime: 1000,
endTime: 2000,
},
],
});
// Mock generateTraceContextIfNeeded
vi.mocked(evaluatorTracing.generateTraceContextIfNeeded).mockResolvedValue({
traceparent: `00-${testTraceId}-0123456789abcdef-01`,
evaluationId: 'test-eval-id',
testCaseId: 'test-case-id',
});
const testSuite: TestSuite = {
providers: [
createMockProvider({
id: 'mock-provider',
response: { output: 'Test response', tokenUsage: {} },
}),
],
prompts: [{ raw: 'Test prompt', label: 'test' }],
tests: [
{
vars: { input: 'test' },
metadata: {
tracingEnabled: true,
evaluationId: 'test-eval-id',
},
assert: [
{
type: 'javascript',
value: `
// Verify trace data is available
if (!context.trace) return false;
return context.trace.spans.length > 0 &&
context.trace.spans[0].name === 'test.operation';
`,
},
],
},
],
tracing: {
enabled: true,
otlp: {
http: {
enabled: true,
port: 4318,
host: '0.0.0.0',
acceptFormats: ['protobuf'],
},
},
},
};
const options: EvaluateOptions = {
maxConcurrency: 1,
};
// Run evaluation
await evaluate(testSuite, mockEval, options);
// Verify trace context was generated
expect(evaluatorTracing.generateTraceContextIfNeeded).toHaveBeenCalled();
expect(evaluatorTracing.startOtlpReceiverIfNeeded).toHaveBeenCalledWith(
testSuite,
'test-eval-id',
);
// Verify trace was fetched for assertion
expect(mockTraceStore.getTrace).toHaveBeenCalledWith(testTraceId, {
sanitizeAttributes: false,
});
expect(mockFlushOtel).toHaveBeenCalled();
expect(mockShutdownOtel).toHaveBeenCalledOnce();
// Verify result was added with passing assertion
expect(mockEval.addResult).toHaveBeenCalledWith(
expect.objectContaining({
success: true,
score: 1,
}),
);
});
it('closes writers but skips OTEL shutdown when initialization fails', async () => {
const writer = {
write: vi.fn().mockResolvedValue(undefined),
close: vi.fn().mockResolvedValue(undefined),
};
const runtime: EvaluatorRuntime<Eval, EvalResult> = {
createEvaluationStore: nodeEvaluatorRuntime.createEvaluationStore,
createResultWriters: vi.fn().mockReturnValue([writer]),
};
mockInitializeOtel.mockImplementationOnce(() => {
throw new Error('otel unavailable');
});
await expect(
evaluate(
{
providers: [],
prompts: [],
tests: [],
tracing: { enabled: true },
},
mockEval,
{},
runtime,
),
).rejects.toThrow('otel unavailable');
expect(writer.close).toHaveBeenCalledOnce();
expect(mockFlushOtel).not.toHaveBeenCalled();
expect(mockShutdownOtel).not.toHaveBeenCalled();
});
it('flushes and shuts down OTEL after successful tracing initialization', async () => {
await evaluate(
{
providers: [
createMockProvider({
id: 'mock-provider',
response: { output: 'Test response', tokenUsage: {} },
}),
],
prompts: [{ raw: 'Test prompt', label: 'test' }],
tests: [{}],
tracing: { enabled: true },
},
mockEval,
{},
);
expect(mockInitializeOtel).toHaveBeenCalledOnce();
expect(mockFlushOtel).toHaveBeenCalledOnce();
expect(mockShutdownOtel).toHaveBeenCalledOnce();
});
it('should handle assertions gracefully when tracing is disabled', async () => {
// Mock generateTraceContextIfNeeded to return null when tracing is disabled
vi.mocked(evaluatorTracing.generateTraceContextIfNeeded).mockResolvedValue(null);
const testSuite: TestSuite = {
providers: [
createMockProvider({
id: 'mock-provider',
response: { output: 'Test response', tokenUsage: {} },
}),
],
prompts: [{ raw: 'Test prompt', label: 'test' }],
tests: [
{
vars: { input: 'test' },
// No tracingEnabled in metadata
assert: [
{
type: 'javascript',
value: `
// Should pass when trace is undefined
return context.trace === undefined && output === 'Test response';
`,
},
],
},
],
// Tracing not enabled in test suite
};
const options: EvaluateOptions = {
maxConcurrency: 1,
};
// Run evaluation
await evaluate(testSuite, mockEval, options);
// Verify trace was NOT created or fetched
expect(mockTraceStore.createTrace).not.toHaveBeenCalled();
expect(mockTraceStore.getTrace).not.toHaveBeenCalled();
expect(mockFlushOtel).not.toHaveBeenCalled();
expect(mockShutdownOtel).not.toHaveBeenCalled();
// Verify result was added with passing assertion
expect(mockEval.addResult).toHaveBeenCalledWith(
expect.objectContaining({
success: true,
score: 1,
}),
);
});
it('should extract traceId correctly from traceparent header', async () => {
const testTraceId = '0af7651916cd43dd8448eb211c80319c';
const testSpanId = 'b7ad6b7169203331';
// Mock the trace context generation
vi.mocked(evaluatorTracing.generateTraceContextIfNeeded).mockResolvedValue({
traceparent: `00-${testTraceId}-${testSpanId}-01`,
evaluationId: 'test-eval-id',
testCaseId: 'test-case-id',
});
mockTraceStore.createTrace.mockResolvedValue(undefined);
mockTraceStore.getTrace.mockResolvedValue({
traceId: testTraceId,
spans: [
{
spanId: 'test-span',
name: 'extracted.correctly',
startTime: 1000,
endTime: 2000,
},
],
});
const testSuite: TestSuite = {
providers: [
createMockProvider({
id: 'mock-provider',
response: { output: 'Test response', tokenUsage: {} },
}),
],
prompts: [{ raw: 'Test prompt', label: 'test' }],
tests: [
{
vars: { input: 'test' },
metadata: {
tracingEnabled: true,
evaluationId: 'test-eval-id',
},
assert: [
{
type: 'javascript',
value: `
// Verify the extracted traceId matches
return context.trace && context.trace.traceId === '${testTraceId}';
`,
},
],
},
],
};
const options: EvaluateOptions = {
maxConcurrency: 1,
};
// Run evaluation
await evaluate(testSuite, mockEval, options);
// Verify trace was fetched with the correct traceId
expect(mockTraceStore.getTrace).toHaveBeenCalledWith(testTraceId, {
sanitizeAttributes: false,
});
// Verify result was added with passing assertion
expect(mockEval.addResult).toHaveBeenCalledWith(
expect.objectContaining({
success: true,
}),
);
});
it('should still run evaluator cleanup when receiver startup is fatal', async () => {
vi.mocked(evaluatorTracing.startOtlpReceiverIfNeeded).mockRejectedValueOnce(
new Error('receiver failed'),
);
const testSuite: TestSuite = {
providers: [],
prompts: [],
tests: [],
tracing: {
enabled: true,
failOnReceiverStartFailure: true,
otlp: {
http: {
enabled: true,
port: 4318,
host: '127.0.0.1',
},
},
},
};
await expect(evaluate(testSuite, mockEval, {})).rejects.toThrow('receiver failed');
expect(evaluatorTracing.stopOtlpReceiverIfNeeded).toHaveBeenCalled();
expect(mockFlushOtel).not.toHaveBeenCalled();
expect(mockShutdownOtel).not.toHaveBeenCalled();
});
});