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 +1,348 @@
|
||||
import { describe, expect, it } from 'vitest';
|
||||
import { handleTraceSpanCount } from '../../src/assertions/traceSpanCount';
|
||||
import { createMockProvider, createProviderResponse } from '../factories/provider';
|
||||
|
||||
import type { AssertionParams, AtomicTestCase } from '../../src/types/index';
|
||||
import type { TraceData } from '../../src/types/tracing';
|
||||
|
||||
const mockProvider = createMockProvider({
|
||||
id: 'mock',
|
||||
response: createProviderResponse({ output: 'mock' }),
|
||||
});
|
||||
|
||||
const mockTraceData: TraceData = {
|
||||
traceId: 'test-trace-id',
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
spans: [
|
||||
{
|
||||
spanId: 'span-1',
|
||||
name: 'llm.completion',
|
||||
startTime: 1000,
|
||||
endTime: 1500,
|
||||
},
|
||||
{
|
||||
spanId: 'span-2',
|
||||
name: 'llm.chat',
|
||||
startTime: 1600,
|
||||
endTime: 2000,
|
||||
},
|
||||
{
|
||||
spanId: 'span-3',
|
||||
name: 'database.query',
|
||||
startTime: 2100,
|
||||
endTime: 2200,
|
||||
},
|
||||
{
|
||||
spanId: 'span-4',
|
||||
name: 'retrieval.search',
|
||||
startTime: 2300,
|
||||
endTime: 2400,
|
||||
},
|
||||
{
|
||||
spanId: 'span-5',
|
||||
name: 'api.external_call',
|
||||
startTime: 2500,
|
||||
endTime: 2600,
|
||||
},
|
||||
],
|
||||
};
|
||||
|
||||
const defaultParams = {
|
||||
baseType: 'trace-span-count' as const,
|
||||
assertionValueContext: {
|
||||
vars: {},
|
||||
test: {} as AtomicTestCase,
|
||||
prompt: 'test prompt',
|
||||
logProbs: undefined,
|
||||
provider: mockProvider,
|
||||
providerResponse: { output: 'test output' },
|
||||
},
|
||||
output: 'test output',
|
||||
outputString: 'test output',
|
||||
providerResponse: { output: 'test output' },
|
||||
test: {} as AtomicTestCase,
|
||||
inverse: false,
|
||||
};
|
||||
|
||||
describe('handleTraceSpanCount', () => {
|
||||
it('should pass when span count is within max limit', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*llm*', max: 3 },
|
||||
},
|
||||
renderedValue: { pattern: '*llm*', max: 3 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
...mockTraceData,
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: true,
|
||||
score: 1,
|
||||
reason:
|
||||
'Found 2 spans matching pattern "*llm*" (expected at most 3). Matched spans: llm.completion, llm.chat',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should fail when span count exceeds max limit', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*llm*', max: 1 },
|
||||
},
|
||||
renderedValue: { pattern: '*llm*', max: 1 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
...mockTraceData,
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: false,
|
||||
score: 0,
|
||||
reason:
|
||||
'Found 2 spans matching pattern "*llm*", expected at most 1. Matched spans: llm.completion, llm.chat',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should pass when span count meets min requirement', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*retrieval*', min: 1 },
|
||||
},
|
||||
renderedValue: { pattern: '*retrieval*', min: 1 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
...mockTraceData,
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: true,
|
||||
score: 1,
|
||||
reason:
|
||||
'Found 1 spans matching pattern "*retrieval*" (expected at least 1). Matched spans: retrieval.search',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should fail when span count is below min requirement', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*retrieval*', min: 2 },
|
||||
},
|
||||
renderedValue: { pattern: '*retrieval*', min: 2 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
...mockTraceData,
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: false,
|
||||
score: 0,
|
||||
reason:
|
||||
'Found 1 spans matching pattern "*retrieval*", expected at least 2. Matched spans: retrieval.search',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should pass when span count is within min and max range', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*', min: 3, max: 10 },
|
||||
},
|
||||
renderedValue: { pattern: '*', min: 3, max: 10 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
...mockTraceData,
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: true,
|
||||
score: 1,
|
||||
reason:
|
||||
'Found 5 spans matching pattern "*" (expected 3-10). Matched spans: llm.completion, llm.chat, database.query, retrieval.search, api.external_call',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should handle patterns with ? wildcard', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: 'llm.c?at', max: 1 },
|
||||
},
|
||||
renderedValue: { pattern: 'llm.c?at', max: 1 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: mockTraceData,
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: true,
|
||||
score: 1,
|
||||
reason:
|
||||
'Found 1 spans matching pattern "llm.c?at" (expected at most 1). Matched spans: llm.chat',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should handle case-insensitive matching', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*LLM*', max: 5 },
|
||||
},
|
||||
renderedValue: { pattern: '*LLM*', max: 5 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
...mockTraceData,
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: true,
|
||||
score: 1,
|
||||
reason:
|
||||
'Found 2 spans matching pattern "*LLM*" (expected at most 5). Matched spans: llm.completion, llm.chat',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should fail when no trace data is available', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*', min: 1 },
|
||||
},
|
||||
renderedValue: { pattern: '*', min: 1 },
|
||||
};
|
||||
|
||||
expect(() => handleTraceSpanCount(params)).toThrow(
|
||||
'No trace data available for trace-span-count assertion',
|
||||
);
|
||||
});
|
||||
|
||||
it('should handle empty trace spans', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: {
|
||||
type: 'trace-span-count',
|
||||
value: { pattern: '*', max: 0 },
|
||||
},
|
||||
renderedValue: { pattern: '*', max: 0 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
traceId: 'empty-trace',
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
spans: [],
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const result = handleTraceSpanCount(params);
|
||||
expect(result).toEqual({
|
||||
pass: true,
|
||||
score: 1,
|
||||
reason: 'Found 0 spans matching pattern "*" (expected at most 0)',
|
||||
assertion: params.assertion,
|
||||
});
|
||||
});
|
||||
|
||||
it('should throw error for invalid value format', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: { type: 'trace-span-count', value: 'invalid' },
|
||||
renderedValue: 'invalid',
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: mockTraceData,
|
||||
},
|
||||
};
|
||||
|
||||
expect(() => handleTraceSpanCount(params)).toThrow(
|
||||
'trace-span-count assertion must have a value object with pattern property',
|
||||
);
|
||||
});
|
||||
|
||||
it('should throw error for missing pattern', () => {
|
||||
const params: AssertionParams = {
|
||||
...defaultParams,
|
||||
assertion: { type: 'trace-span-count', value: { max: 5 } },
|
||||
renderedValue: { max: 5 },
|
||||
assertionValueContext: {
|
||||
...defaultParams.assertionValueContext,
|
||||
trace: {
|
||||
...mockTraceData,
|
||||
evaluationId: 'test-evaluation-id',
|
||||
testCaseId: 'test-test-case-id',
|
||||
metadata: { test: 'value' },
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
expect(() => handleTraceSpanCount(params)).toThrow(
|
||||
'trace-span-count assertion must have a value object with pattern property',
|
||||
);
|
||||
});
|
||||
});
|
||||
Reference in New Issue
Block a user