Files
promptfoo--promptfoo/test/redteam/plugins/harmful/graders.test.ts
T
wehub-resource-sync 0d3cb498a3
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
chore: import upstream snapshot with attribution
2026-07-13 13:24:08 +08:00

326 lines
8.5 KiB
TypeScript

import { beforeEach, describe, expect, it, vi } from 'vitest';
import {
CopyrightViolationGrader,
GraphicContentGrader,
HarmfulGrader,
HarmfulPrivacyGrader,
ProfanityGrader,
SelfHarmGrader,
SexualContentGrader,
} from '../../../../src/redteam/plugins/harmful/graders';
import { isBasicRefusal } from '../../../../src/redteam/util';
import { createMockProvider } from '../../../factories/provider';
import type { ApiProvider, AtomicTestCase } from '../../../../src/types/index';
vi.mock('../../../../src/redteam/util');
const createMockTest = (overrides?: Partial<AtomicTestCase>): AtomicTestCase => ({
vars: {
purpose: 'To assist users with general tasks while avoiding harmful content',
harmCategory: 'test harm',
prompt: 'test prompt',
entities: ['entity1', 'entity2'],
...overrides?.vars,
},
...overrides,
});
describe('HarmfulGrader', () => {
let grader: HarmfulGrader;
let mockProvider: Partial<ApiProvider>;
let mockTest: AtomicTestCase;
beforeEach(() => {
grader = new HarmfulGrader();
mockProvider = createMockProvider();
mockTest = createMockTest();
vi.mocked(isBasicRefusal).mockReset();
});
describe('getResult', () => {
it('should pass and return score 1 for basic refusal', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return true;
});
const result = await grader.getResult(
'test prompt',
'I cannot help with that.',
mockTest,
mockProvider as ApiProvider,
);
expect(result.grade).toEqual({
pass: true,
score: 1,
reason: 'The output contains a standard refusal.',
assertion: undefined,
});
expect(result.rubric).toBe(grader.rubric);
});
it('should fallback to superclass getResult for non-refusal', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return false;
});
const superGetResult = vi.spyOn(Object.getPrototypeOf(HarmfulGrader.prototype), 'getResult');
superGetResult.mockResolvedValue({
grade: {
pass: false,
score: 0,
reason: 'test reason',
},
rubric: 'test rubric',
});
const result = await grader.getResult(
'test prompt',
'harmful response',
mockTest,
mockProvider as ApiProvider,
);
expect(superGetResult).toHaveBeenCalledWith(
'test prompt',
'harmful response',
mockTest,
mockProvider as ApiProvider,
undefined,
undefined, // additionalRubric
undefined, // skipRefusalCheck
undefined, // gradingContext
);
expect(result).toEqual({
grade: {
pass: false,
score: 0,
reason: 'test reason',
},
rubric: 'test rubric',
});
superGetResult.mockRestore();
});
});
describe('renderRubric', () => {
it('should render rubric with variables', () => {
const vars = {
purpose: 'test purpose',
harmCategory: 'test harm',
prompt: 'test prompt',
entities: ['entity1', 'entity2'],
};
const renderedRubric = grader.renderRubric(vars);
expect(renderedRubric).toContain('test purpose');
expect(renderedRubric).toContain('test harm');
expect(renderedRubric).toContain('test prompt');
expect(renderedRubric).toContain('entity1');
expect(renderedRubric).toContain('entity2');
});
it('should handle missing entities in rubric', () => {
const vars = {
purpose: 'test purpose',
harmCategory: 'test harm',
prompt: 'test prompt',
};
const renderedRubric = grader.renderRubric(vars);
expect(renderedRubric).toContain('test purpose');
expect(renderedRubric).toContain('test harm');
expect(renderedRubric).not.toContain('AllowedEntities');
});
});
});
describe('HarmfulPrivacyGrader', () => {
let grader: HarmfulPrivacyGrader;
let mockTest: AtomicTestCase;
beforeEach(() => {
grader = new HarmfulPrivacyGrader();
mockTest = createMockTest();
});
it('should have privacy-specific rubric', () => {
expect(grader.rubric).toContain('privacy violation');
});
it('should inherit basic refusal handling from parent', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return true;
});
const result = await grader.getResult(
'test prompt',
'I cannot help with that.',
mockTest,
createMockProvider() as ApiProvider,
);
expect(result.grade.pass).toBe(true);
expect(result.grade.score).toBe(1);
});
});
describe('SelfHarmGrader', () => {
let grader: SelfHarmGrader;
let mockTest: AtomicTestCase;
beforeEach(() => {
grader = new SelfHarmGrader();
mockTest = createMockTest();
});
it('should have self-harm specific rubric', () => {
expect(grader.rubric).toContain('self-harm');
});
it('should inherit basic refusal handling from parent', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return true;
});
const result = await grader.getResult(
'test prompt',
'I cannot help with that.',
mockTest,
createMockProvider() as ApiProvider,
);
expect(result.grade.pass).toBe(true);
expect(result.grade.score).toBe(1);
});
});
describe('GraphicContentGrader', () => {
let grader: GraphicContentGrader;
let mockTest: AtomicTestCase;
beforeEach(() => {
grader = new GraphicContentGrader();
mockTest = createMockTest();
});
it('should have graphic content specific rubric', () => {
expect(grader.rubric).toContain('gratuitously graphic');
expect(grader.rubric).toContain('disturbing content');
});
it('should inherit basic refusal handling from parent', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return true;
});
const result = await grader.getResult(
'test prompt',
'I cannot help with that.',
mockTest,
createMockProvider() as ApiProvider,
);
expect(result.grade.pass).toBe(true);
expect(result.grade.score).toBe(1);
});
});
describe('SexualContentGrader', () => {
let grader: SexualContentGrader;
let mockTest: AtomicTestCase;
beforeEach(() => {
grader = new SexualContentGrader();
mockTest = createMockTest();
});
it('should have sexual content specific rubric', () => {
expect(grader.rubric).toContain('inappropriate sexual content');
expect(grader.rubric).toContain('Sexually suggestive');
});
it('should inherit basic refusal handling from parent', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return true;
});
const result = await grader.getResult(
'test prompt',
'I cannot help with that.',
mockTest,
createMockProvider() as ApiProvider,
);
expect(result.grade.pass).toBe(true);
expect(result.grade.score).toBe(1);
});
});
describe('ProfanityGrader', () => {
let grader: ProfanityGrader;
let mockTest: AtomicTestCase;
beforeEach(() => {
grader = new ProfanityGrader();
mockTest = createMockTest();
});
it('should have profanity specific rubric', () => {
expect(grader.rubric).toContain('profanity');
expect(grader.rubric).toContain('inappropriate language');
});
it('should inherit basic refusal handling from parent', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return true;
});
const result = await grader.getResult(
'test prompt',
'I cannot help with that.',
mockTest,
createMockProvider() as ApiProvider,
);
expect(result.grade.pass).toBe(true);
expect(result.grade.score).toBe(1);
});
});
describe('CopyrightViolationGrader', () => {
let grader: CopyrightViolationGrader;
let mockTest: AtomicTestCase;
beforeEach(() => {
grader = new CopyrightViolationGrader();
mockTest = createMockTest();
});
it('should have copyright specific rubric', () => {
expect(grader.rubric).toContain('copyright violations');
expect(grader.rubric).toContain('copyrighted content');
});
it('should inherit basic refusal handling from parent', async () => {
vi.mocked(isBasicRefusal).mockImplementation(function () {
return true;
});
const result = await grader.getResult(
'test prompt',
'I cannot help with that.',
mockTest,
createMockProvider() as ApiProvider,
);
expect(result.grade.pass).toBe(true);
expect(result.grade.score).toBe(1);
});
});