chore: import upstream snapshot with attribution
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
Deploy local.promptfoo.app / Deploy to Cloudflare Pages (push) Has been cancelled
Test and Publish Multi-arch Docker Image / test (push) Has been cancelled
Test and Publish Multi-arch Docker Image / build-docker-and-push-digests (map[digest-suffix:linux-amd64 platform:linux/amd64 runner:ubuntu-latest]) (push) Has been cancelled
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) Has been cancelled
Test and Publish Multi-arch Docker Image / merge-docker-digests (push) Has been cancelled
Test and Publish Multi-arch Docker Image / Attest Multi-arch Image (push) Has been cancelled
Validate Renovate Config / Validate Renovate Configuration (push) Has been cancelled
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
Deploy local.promptfoo.app / Deploy to Cloudflare Pages (push) Has been cancelled
Test and Publish Multi-arch Docker Image / test (push) Has been cancelled
Test and Publish Multi-arch Docker Image / build-docker-and-push-digests (map[digest-suffix:linux-amd64 platform:linux/amd64 runner:ubuntu-latest]) (push) Has been cancelled
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) Has been cancelled
Test and Publish Multi-arch Docker Image / merge-docker-digests (push) Has been cancelled
Test and Publish Multi-arch Docker Image / Attest Multi-arch Image (push) Has been cancelled
Validate Renovate Config / Validate Renovate Configuration (push) Has been cancelled
This commit is contained in:
@@ -0,0 +1,317 @@
|
||||
import { afterAll, beforeAll, beforeEach, describe, expect, it, vi } from 'vitest';
|
||||
import { fetchWithCache } from '../../../src/cache';
|
||||
import { VERSION } from '../../../src/constants';
|
||||
import logger from '../../../src/logger';
|
||||
import { getRequestTimeoutMs } from '../../../src/providers/shared';
|
||||
import {
|
||||
callExtraction,
|
||||
fetchRemoteGeneration,
|
||||
formatPrompts,
|
||||
RedTeamGenerationResponse,
|
||||
} from '../../../src/redteam/extraction/util';
|
||||
import { getRemoteGenerationUrl } from '../../../src/redteam/remoteGeneration';
|
||||
import {
|
||||
createMockProvider,
|
||||
createProviderResponse,
|
||||
type MockApiProvider,
|
||||
} from '../../factories/provider';
|
||||
import { mockProcessEnv } from '../../util/utils';
|
||||
|
||||
vi.mock('../../../src/cache', async (importOriginal) => {
|
||||
return {
|
||||
...(await importOriginal()),
|
||||
fetchWithCache: vi.fn(),
|
||||
};
|
||||
});
|
||||
|
||||
vi.mock('../../../src/logger', () => ({
|
||||
default: {
|
||||
debug: vi.fn(),
|
||||
info: vi.fn(),
|
||||
warn: vi.fn(),
|
||||
error: vi.fn(),
|
||||
},
|
||||
getLogLevel: vi.fn().mockReturnValue('info'),
|
||||
}));
|
||||
|
||||
vi.mock('../../../src/redteam/remoteGeneration', async (importOriginal) => {
|
||||
return {
|
||||
...(await importOriginal()),
|
||||
getRemoteGenerationUrl: vi.fn().mockReturnValue('https://api.promptfoo.app/api/v1/task'),
|
||||
};
|
||||
});
|
||||
|
||||
describe('fetchRemoteGeneration', () => {
|
||||
let restoreEnv: () => void;
|
||||
|
||||
beforeAll(() => {
|
||||
restoreEnv = mockProcessEnv({ PROMPTFOO_REMOTE_GENERATION_URL: undefined });
|
||||
});
|
||||
|
||||
afterAll(() => {
|
||||
restoreEnv();
|
||||
});
|
||||
|
||||
beforeEach(() => {
|
||||
vi.clearAllMocks();
|
||||
vi.mocked(getRemoteGenerationUrl).mockImplementation(function () {
|
||||
return 'https://api.promptfoo.app/api/v1/task';
|
||||
});
|
||||
});
|
||||
|
||||
it('should fetch remote generation for purpose task', async () => {
|
||||
const mockResponse = {
|
||||
data: {
|
||||
task: 'purpose',
|
||||
result: 'This is a purpose',
|
||||
},
|
||||
status: 200,
|
||||
statusText: 'OK',
|
||||
cached: false,
|
||||
};
|
||||
vi.mocked(fetchWithCache).mockResolvedValue(mockResponse);
|
||||
|
||||
const result = await fetchRemoteGeneration('purpose', ['prompt1', 'prompt2']);
|
||||
|
||||
expect(result).toBe('This is a purpose');
|
||||
expect(fetchWithCache).toHaveBeenCalledWith(
|
||||
'https://api.promptfoo.app/api/v1/task',
|
||||
{
|
||||
method: 'POST',
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
body: JSON.stringify({
|
||||
task: 'purpose',
|
||||
prompts: ['prompt1', 'prompt2'],
|
||||
version: VERSION,
|
||||
email: null,
|
||||
}),
|
||||
},
|
||||
getRequestTimeoutMs(),
|
||||
'json',
|
||||
);
|
||||
});
|
||||
|
||||
it('should fetch remote generation for entities task', async () => {
|
||||
const mockResponse = {
|
||||
data: {
|
||||
task: 'entities',
|
||||
result: ['Entity1', 'Entity2'],
|
||||
},
|
||||
status: 200,
|
||||
statusText: 'OK',
|
||||
cached: false,
|
||||
};
|
||||
vi.mocked(fetchWithCache).mockResolvedValue(mockResponse);
|
||||
|
||||
const result = await fetchRemoteGeneration('entities', ['prompt1', 'prompt2']);
|
||||
|
||||
expect(result).toEqual(['Entity1', 'Entity2']);
|
||||
expect(fetchWithCache).toHaveBeenCalledWith(
|
||||
'https://api.promptfoo.app/api/v1/task',
|
||||
{
|
||||
method: 'POST',
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
body: JSON.stringify({
|
||||
task: 'entities',
|
||||
prompts: ['prompt1', 'prompt2'],
|
||||
version: VERSION,
|
||||
email: null,
|
||||
}),
|
||||
},
|
||||
getRequestTimeoutMs(),
|
||||
'json',
|
||||
);
|
||||
});
|
||||
|
||||
it('should include the resolved cloud target in the remote generation payload', async () => {
|
||||
vi.mocked(fetchWithCache).mockResolvedValue({
|
||||
data: { task: 'purpose', result: 'This is a purpose' },
|
||||
status: 200,
|
||||
statusText: 'OK',
|
||||
cached: false,
|
||||
});
|
||||
|
||||
await fetchRemoteGeneration('purpose', ['prompt'], {
|
||||
providerTargetIds: ['file://local-provider.ts'],
|
||||
cloudTargetId: 'cloud-target-123',
|
||||
});
|
||||
|
||||
expect(fetchWithCache).toHaveBeenCalledWith(
|
||||
'https://api.promptfoo.app/api/v1/task',
|
||||
expect.objectContaining({
|
||||
body: JSON.stringify({
|
||||
task: 'purpose',
|
||||
prompts: ['prompt'],
|
||||
version: VERSION,
|
||||
email: null,
|
||||
targetId: 'cloud-target-123',
|
||||
}),
|
||||
}),
|
||||
getRequestTimeoutMs(),
|
||||
'json',
|
||||
);
|
||||
});
|
||||
|
||||
it('should throw an error when fetchWithCache fails', async () => {
|
||||
const mockError = new Error('Network error');
|
||||
vi.mocked(fetchWithCache).mockRejectedValue(mockError);
|
||||
|
||||
await expect(fetchRemoteGeneration('purpose', ['prompt'])).rejects.toThrow('Network error');
|
||||
expect(logger.warn).toHaveBeenCalledWith(
|
||||
"Error using remote generation for task 'purpose': Error: Network error",
|
||||
);
|
||||
});
|
||||
|
||||
it('should throw an error when response parsing fails', async () => {
|
||||
const mockResponse = {
|
||||
data: {
|
||||
task: 'purpose',
|
||||
// Missing 'result' field
|
||||
},
|
||||
status: 200,
|
||||
statusText: 'OK',
|
||||
cached: false,
|
||||
};
|
||||
vi.mocked(fetchWithCache).mockResolvedValue(mockResponse);
|
||||
|
||||
await expect(fetchRemoteGeneration('purpose', ['prompt'])).rejects.toThrow('Invalid input');
|
||||
expect(logger.warn).toHaveBeenCalledWith(
|
||||
expect.stringContaining("Error using remote generation for task 'purpose':"),
|
||||
);
|
||||
});
|
||||
|
||||
it('should use custom remote generation URL when provided', async () => {
|
||||
const customUrl = 'https://custom-api.example.com/generate';
|
||||
vi.mocked(getRemoteGenerationUrl).mockImplementation(function () {
|
||||
return customUrl;
|
||||
});
|
||||
|
||||
const mockResponse = {
|
||||
data: {
|
||||
task: 'purpose',
|
||||
result: 'This is a purpose',
|
||||
},
|
||||
status: 200,
|
||||
statusText: 'OK',
|
||||
cached: false,
|
||||
};
|
||||
vi.mocked(fetchWithCache).mockResolvedValue(mockResponse);
|
||||
|
||||
await fetchRemoteGeneration('purpose', ['prompt1']);
|
||||
|
||||
expect(fetchWithCache).toHaveBeenCalledWith(
|
||||
customUrl,
|
||||
expect.any(Object),
|
||||
getRequestTimeoutMs(),
|
||||
'json',
|
||||
);
|
||||
});
|
||||
});
|
||||
|
||||
describe('RedTeamGenerationResponse', () => {
|
||||
it('should validate correct response structure', () => {
|
||||
const validResponse = {
|
||||
task: 'purpose',
|
||||
result: 'This is a purpose',
|
||||
};
|
||||
|
||||
expect(() => RedTeamGenerationResponse.parse(validResponse)).not.toThrow();
|
||||
});
|
||||
|
||||
it('should throw error for invalid response structure', () => {
|
||||
const invalidResponse = {
|
||||
task: 'purpose',
|
||||
// Missing 'result' field
|
||||
};
|
||||
|
||||
expect(() => RedTeamGenerationResponse.parse(invalidResponse)).toThrow('Invalid input');
|
||||
});
|
||||
|
||||
it('should validate response with string result', () => {
|
||||
const response = {
|
||||
task: 'purpose',
|
||||
result: 'This is a purpose',
|
||||
};
|
||||
|
||||
expect(() => RedTeamGenerationResponse.parse(response)).not.toThrow();
|
||||
});
|
||||
|
||||
it('should validate response with array result', () => {
|
||||
const response = {
|
||||
task: 'entities',
|
||||
result: ['Entity1', 'Entity2'],
|
||||
};
|
||||
|
||||
expect(() => RedTeamGenerationResponse.parse(response)).not.toThrow();
|
||||
});
|
||||
});
|
||||
|
||||
describe('Extraction Utils', () => {
|
||||
let provider: MockApiProvider;
|
||||
|
||||
beforeEach(() => {
|
||||
provider = createMockProvider({
|
||||
response: createProviderResponse({ output: 'test output' }),
|
||||
});
|
||||
vi.clearAllMocks();
|
||||
});
|
||||
|
||||
describe('callExtraction', () => {
|
||||
it('should call API with formatted chat message and process output correctly', async () => {
|
||||
const result = await callExtraction(provider, 'test prompt', (output) =>
|
||||
output.toUpperCase(),
|
||||
);
|
||||
expect(result).toBe('TEST OUTPUT');
|
||||
expect(provider.callApi).toHaveBeenCalledWith(
|
||||
JSON.stringify([{ role: 'user', content: 'test prompt' }]),
|
||||
);
|
||||
});
|
||||
|
||||
it('should throw an error if API call fails', async () => {
|
||||
const error = new Error('API error');
|
||||
vi.mocked(provider.callApi).mockResolvedValue({ error: error.message });
|
||||
|
||||
await expect(callExtraction(provider, 'test prompt', vi.fn())).rejects.toThrow(
|
||||
'Failed to perform extraction: API error',
|
||||
);
|
||||
});
|
||||
|
||||
it('should throw an error if output is not a string', async () => {
|
||||
vi.mocked(provider.callApi).mockResolvedValue({ output: 123 });
|
||||
|
||||
await expect(callExtraction(provider, 'test prompt', vi.fn())).rejects.toThrow(
|
||||
'Invalid extraction output: expected string, got: 123',
|
||||
);
|
||||
});
|
||||
|
||||
it('should handle empty string output', async () => {
|
||||
vi.mocked(provider.callApi).mockResolvedValue({ output: '' });
|
||||
|
||||
const result = await callExtraction(provider, 'test prompt', (output) => output.length);
|
||||
expect(result).toBe(0);
|
||||
});
|
||||
|
||||
it('should handle null output', async () => {
|
||||
vi.mocked(provider.callApi).mockResolvedValue({ output: null });
|
||||
|
||||
await expect(callExtraction(provider, 'test prompt', vi.fn())).rejects.toThrow(
|
||||
'Invalid extraction output: expected string, got: null',
|
||||
);
|
||||
});
|
||||
|
||||
it('should handle undefined output', async () => {
|
||||
vi.mocked(provider.callApi).mockResolvedValue({ output: undefined });
|
||||
|
||||
await expect(callExtraction(provider, 'test prompt', vi.fn())).rejects.toThrow(
|
||||
'Invalid extraction output: expected string, got: undefined',
|
||||
);
|
||||
});
|
||||
});
|
||||
|
||||
describe('formatPrompts', () => {
|
||||
it('should format prompts correctly', () => {
|
||||
const formattedPrompts = formatPrompts(['prompt1', 'prompt2']);
|
||||
expect(formattedPrompts).toBe('<Prompt>\nprompt1\n</Prompt>\n<Prompt>\nprompt2\n</Prompt>');
|
||||
});
|
||||
});
|
||||
});
|
||||
Reference in New Issue
Block a user