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 +1,323 @@
|
||||
import { describe, expect, it } from 'vitest';
|
||||
import { handleLevenshtein } from '../../src/assertions/levenshtein';
|
||||
|
||||
describe('handleLevenshtein', () => {
|
||||
it('should pass when strings are identical', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein' },
|
||||
renderedValue: 'test',
|
||||
outputString: 'test',
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: 'test',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'test', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: 'test',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(true);
|
||||
expect(result.score).toBe(1);
|
||||
expect(result.reason).toBe('Assertion passed');
|
||||
});
|
||||
|
||||
it('should pass when distance is within default threshold', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein' },
|
||||
renderedValue: 'test',
|
||||
outputString: 'tast', // Distance of 1
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: 'tast',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'tast', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: 'tast',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(true);
|
||||
expect(result.score).toBe(1);
|
||||
});
|
||||
|
||||
it('should pass when distance is within custom threshold', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein', threshold: 2 },
|
||||
renderedValue: 'test',
|
||||
outputString: 'tost', // Distance of 1
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: 'tost',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'tost', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: 'tost',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(true);
|
||||
expect(result.score).toBe(1);
|
||||
});
|
||||
|
||||
it('should fail when distance exceeds threshold', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein', threshold: 1 },
|
||||
renderedValue: 'test',
|
||||
outputString: 'toast', // Distance of 2
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: 'toast',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'toast', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: 'toast',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(false);
|
||||
expect(result.score).toBe(0);
|
||||
expect(result.reason).toBe('Levenshtein distance 2 is greater than threshold 1');
|
||||
});
|
||||
|
||||
it('should fail when distance exceeds default threshold', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein' },
|
||||
renderedValue: 'test',
|
||||
outputString: 'completely different', // Distance > 5
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: 'completely different',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'completely different', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: 'completely different',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(false);
|
||||
expect(result.score).toBe(0);
|
||||
});
|
||||
|
||||
it('should throw error when renderedValue is not a string', () => {
|
||||
expect(() =>
|
||||
handleLevenshtein({
|
||||
assertion: { type: 'levenshtein' },
|
||||
renderedValue: 123 as any,
|
||||
outputString: 'test',
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: 'test',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'test', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: 'test',
|
||||
}),
|
||||
).toThrow('"levenshtein" assertion type must have a string value');
|
||||
});
|
||||
|
||||
it('should handle empty strings', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein' },
|
||||
renderedValue: '',
|
||||
outputString: '',
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: '',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: '', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: '',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(true);
|
||||
expect(result.score).toBe(1);
|
||||
});
|
||||
|
||||
it('should handle long strings', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein', threshold: 10 },
|
||||
renderedValue: 'this is a very long string to test',
|
||||
outputString: 'this is a vary long string to test',
|
||||
test: {},
|
||||
providerResponse: {
|
||||
output: 'this is a vary long string to test',
|
||||
tokenUsage: {},
|
||||
},
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'this is a vary long string to test', tokenUsage: {} },
|
||||
},
|
||||
inverse: false,
|
||||
output: 'this is a vary long string to test',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(true);
|
||||
expect(result.score).toBe(1);
|
||||
});
|
||||
|
||||
describe('inverse (not-levenshtein)', () => {
|
||||
it('should fail when distance is within threshold', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein', threshold: 5 },
|
||||
renderedValue: 'test',
|
||||
outputString: 'tast', // Distance of 1, within threshold
|
||||
test: {},
|
||||
providerResponse: { output: 'tast', tokenUsage: {} },
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'tast', tokenUsage: {} },
|
||||
},
|
||||
inverse: true,
|
||||
output: 'tast',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(false);
|
||||
expect(result.score).toBe(0);
|
||||
expect(result.reason).toBe('Levenshtein distance 1 is less than or equal to threshold 5');
|
||||
});
|
||||
|
||||
it('should pass when distance exceeds threshold', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein', threshold: 2 },
|
||||
renderedValue: 'test',
|
||||
outputString: 'completely different', // Distance well over threshold
|
||||
test: {},
|
||||
providerResponse: { output: 'completely different', tokenUsage: {} },
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'completely different', tokenUsage: {} },
|
||||
},
|
||||
inverse: true,
|
||||
output: 'completely different',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(true);
|
||||
expect(result.score).toBe(1);
|
||||
expect(result.reason).toBe('Assertion passed');
|
||||
});
|
||||
|
||||
it('should fail at the exact boundary (distance === threshold)', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein', threshold: 2 },
|
||||
renderedValue: 'test',
|
||||
outputString: 'toast', // Distance of exactly 2, equal to threshold
|
||||
test: {},
|
||||
providerResponse: { output: 'toast', tokenUsage: {} },
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'toast', tokenUsage: {} },
|
||||
},
|
||||
inverse: true,
|
||||
output: 'toast',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(false);
|
||||
expect(result.score).toBe(0);
|
||||
expect(result.reason).toBe('Levenshtein distance 2 is less than or equal to threshold 2');
|
||||
});
|
||||
|
||||
it('should honor inverse with the default threshold (no threshold specified)', () => {
|
||||
const result = handleLevenshtein({
|
||||
assertion: { type: 'levenshtein' }, // Default threshold of 5
|
||||
renderedValue: 'test',
|
||||
outputString: 'tast', // Distance of 1, within default threshold
|
||||
test: {},
|
||||
providerResponse: { output: 'tast', tokenUsage: {} },
|
||||
baseType: 'contains' as any,
|
||||
assertionValueContext: {
|
||||
prompt: '',
|
||||
vars: {},
|
||||
test: {},
|
||||
logProbs: undefined,
|
||||
provider: {} as any,
|
||||
providerResponse: { output: 'tast', tokenUsage: {} },
|
||||
},
|
||||
inverse: true,
|
||||
output: 'tast',
|
||||
});
|
||||
|
||||
expect(result.pass).toBe(false);
|
||||
expect(result.score).toBe(0);
|
||||
expect(result.reason).toBe('Levenshtein distance 1 is less than or equal to threshold 5');
|
||||
});
|
||||
});
|
||||
});
|
||||
Reference in New Issue
Block a user