Merge branch 'n8n-io:master' into master

packages/cli/src/evaluation/test-runner/__tests__/evaluation-metrics.ee.test.ts

@@ -0,0 +1,72 @@
+import { EvaluationMetrics } from '../evaluation-metrics.ee';
+
+describe('EvaluationMetrics', () => {
+	test('should aggregate metrics correctly', () => {
+		const testMetricNames = new Set(['metric1', 'metric2']);
+		const metrics = new EvaluationMetrics(testMetricNames);
+
+		metrics.addResults({ metric1: 1, metric2: 0 });
+		metrics.addResults({ metric1: 0.5, metric2: 0.2 });
+
+		const aggregatedMetrics = metrics.getAggregatedMetrics();
+
+		expect(aggregatedMetrics).toEqual({ metric1: 0.75, metric2: 0.1 });
+	});
+
+	test('should aggregate only numbers', () => {
+		const testMetricNames = new Set(['metric1', 'metric2']);
+		const metrics = new EvaluationMetrics(testMetricNames);
+
+		metrics.addResults({ metric1: 1, metric2: 0 });
+		metrics.addResults({ metric1: '0.5', metric2: 0.2 });
+		metrics.addResults({ metric1: 'not a number', metric2: [1, 2, 3] });
+
+		const aggregatedUpMetrics = metrics.getAggregatedMetrics();
+
+		expect(aggregatedUpMetrics).toEqual({ metric1: 1, metric2: 0.1 });
+	});
+
+	test('should handle missing values', () => {
+		const testMetricNames = new Set(['metric1', 'metric2']);
+		const metrics = new EvaluationMetrics(testMetricNames);
+
+		metrics.addResults({ metric1: 1 });
+		metrics.addResults({ metric2: 0.2 });
+
+		const aggregatedMetrics = metrics.getAggregatedMetrics();
+
+		expect(aggregatedMetrics).toEqual({ metric1: 1, metric2: 0.2 });
+	});
+
+	test('should handle empty metrics', () => {
+		const testMetricNames = new Set(['metric1', 'metric2']);
+		const metrics = new EvaluationMetrics(testMetricNames);
+
+		const aggregatedMetrics = metrics.getAggregatedMetrics();
+
+		expect(aggregatedMetrics).toEqual({});
+	});
+
+	test('should handle empty testMetrics', () => {
+		const metrics = new EvaluationMetrics(new Set());
+
+		metrics.addResults({ metric1: 1, metric2: 0 });
+		metrics.addResults({ metric1: 0.5, metric2: 0.2 });
+
+		const aggregatedMetrics = metrics.getAggregatedMetrics();
+
+		expect(aggregatedMetrics).toEqual({});
+	});
+
+	test('should ignore non-relevant values', () => {
+		const testMetricNames = new Set(['metric1']);
+		const metrics = new EvaluationMetrics(testMetricNames);
+
+		metrics.addResults({ metric1: 1, notRelevant: 0 });
+		metrics.addResults({ metric1: 0.5, notRelevant2: { foo: 'bar' } });
+
+		const aggregatedMetrics = metrics.getAggregatedMetrics();
+
+		expect(aggregatedMetrics).toEqual({ metric1: 0.75 });
+	});
+});

packages/cli/src/evaluation/test-runner/__tests__/mock-data/workflow.evaluation.json

@@ -57,6 +57,12 @@
 							"name": "success",
 							"value": true,
 							"type": "boolean"
+						},
+						{
+							"id": "877d1bf8-31a7-4571-9293-a6837b51d22b",
+							"name": "metric1",
+							"value": 0.1,
+							"type": "number"
 						}
 					]
 				},

packages/cli/src/evaluation/test-runner/__tests__/test-runner.service.ee.test.ts

@@ -2,15 +2,17 @@ import type { SelectQueryBuilder } from '@n8n/typeorm';
 import { stringify } from 'flatted';
 import { readFileSync } from 'fs';
 import { mock, mockDeep } from 'jest-mock-extended';
-import type { IRun } from 'n8n-workflow';
+import type { GenericValue, IRun } from 'n8n-workflow';
 import path from 'path';
 
 import type { ActiveExecutions } from '@/active-executions';
 import type { ExecutionEntity } from '@/databases/entities/execution-entity';
 import type { TestDefinition } from '@/databases/entities/test-definition.ee';
+import type { TestMetric } from '@/databases/entities/test-metric.ee';
 import type { TestRun } from '@/databases/entities/test-run.ee';
 import type { User } from '@/databases/entities/user';
 import type { ExecutionRepository } from '@/databases/repositories/execution.repository';
+import type { TestMetricRepository } from '@/databases/repositories/test-metric.repository.ee';
 import type { TestRunRepository } from '@/databases/repositories/test-run.repository.ee';
 import type { WorkflowRepository } from '@/databases/repositories/workflow.repository';
 import type { WorkflowRunner } from '@/workflow-runner';
@@ -58,12 +60,38 @@ function mockExecutionData() {
 	});
 }
 
+function mockEvaluationExecutionData(metrics: Record<string, GenericValue>) {
+	return mock<IRun>({
+		data: {
+			resultData: {
+				lastNodeExecuted: 'lastNode',
+				runData: {
+					lastNode: [
+						{
+							data: {
+								main: [
+									[
+										{
+											json: metrics,
+										},
+									],
+								],
+							},
+						},
+					],
+				},
+			},
+		},
+	});
+}
+
 describe('TestRunnerService', () => {
 	const executionRepository = mock<ExecutionRepository>();
 	const workflowRepository = mock<WorkflowRepository>();
 	const workflowRunner = mock<WorkflowRunner>();
 	const activeExecutions = mock<ActiveExecutions>();
 	const testRunRepository = mock<TestRunRepository>();
+	const testMetricRepository = mock<TestMetricRepository>();
 
 	beforeEach(() => {
 		const executionsQbMock = mockDeep<SelectQueryBuilder<ExecutionEntity>>({
@@ -80,6 +108,11 @@ describe('TestRunnerService', () => {
 			.mockResolvedValueOnce(executionMocks[1]);
 
 		testRunRepository.createTestRun.mockResolvedValue(mock<TestRun>({ id: 'test-run-id' }));
+
+		testMetricRepository.find.mockResolvedValue([
+			mock<TestMetric>({ name: 'metric1' }),
+			mock<TestMetric>({ name: 'metric2' }),
+		]);
 	});
 
 	afterEach(() => {
@@ -97,6 +130,7 @@ describe('TestRunnerService', () => {
 			executionRepository,
 			activeExecutions,
 			testRunRepository,
+			testMetricRepository,
 		);
 
 		expect(testRunnerService).toBeInstanceOf(TestRunnerService);
@@ -109,6 +143,7 @@ describe('TestRunnerService', () => {
 			executionRepository,
 			activeExecutions,
 			testRunRepository,
+			testMetricRepository,
 		);
 
 		workflowRepository.findById.calledWith('workflow-under-test-id').mockResolvedValueOnce({
@@ -143,6 +178,7 @@ describe('TestRunnerService', () => {
 			executionRepository,
 			activeExecutions,
 			testRunRepository,
+			testMetricRepository,
 		);
 
 		workflowRepository.findById.calledWith('workflow-under-test-id').mockResolvedValueOnce({
@@ -166,17 +202,17 @@ describe('TestRunnerService', () => {
 			.mockResolvedValue(mockExecutionData());
 
 		activeExecutions.getPostExecutePromise
-			.calledWith('some-execution-id-2')
+			.calledWith('some-execution-id-3')
 			.mockResolvedValue(mockExecutionData());
 
 		// Mock executions of evaluation workflow
 		activeExecutions.getPostExecutePromise
-			.calledWith('some-execution-id-3')
+			.calledWith('some-execution-id-2')
-			.mockResolvedValue(mockExecutionData());
+			.mockResolvedValue(mockEvaluationExecutionData({ metric1: 1, metric2: 0 }));
 
 		activeExecutions.getPostExecutePromise
 			.calledWith('some-execution-id-4')
-			.mockResolvedValue(mockExecutionData());
+			.mockResolvedValue(mockEvaluationExecutionData({ metric1: 0.5 }));
 
 		await testRunnerService.runTest(
 			mock<User>(),
@@ -225,7 +261,8 @@ describe('TestRunnerService', () => {
 		expect(testRunRepository.markAsRunning).toHaveBeenCalledWith('test-run-id');
 		expect(testRunRepository.markAsCompleted).toHaveBeenCalledTimes(1);
 		expect(testRunRepository.markAsCompleted).toHaveBeenCalledWith('test-run-id', {
-			success: false,
+			metric1: 0.75,
+			metric2: 0,
 		});
 	});
 });

packages/cli/src/evaluation/test-runner/evaluation-metrics.ee.ts

@@ -0,0 +1,32 @@
+import type { IDataObject } from 'n8n-workflow';
+
+export class EvaluationMetrics {
+	private readonly rawMetricsByName = new Map<string, number[]>();
+
+	constructor(private readonly metricNames: Set<string>) {
+		for (const metricName of metricNames) {
+			this.rawMetricsByName.set(metricName, []);
+		}
+	}
+
+	addResults(result: IDataObject) {
+		for (const [metricName, metricValue] of Object.entries(result)) {
+			if (typeof metricValue === 'number' && this.metricNames.has(metricName)) {
+				this.rawMetricsByName.get(metricName)!.push(metricValue);
+			}
+		}
+	}
+
+	getAggregatedMetrics() {
+		const aggregatedMetrics: Record<string, number> = {};
+
+		for (const [metricName, metricValues] of this.rawMetricsByName.entries()) {
+			if (metricValues.length > 0) {
+				const metricSum = metricValues.reduce((acc, val) => acc + val, 0);
+				aggregatedMetrics[metricName] = metricSum / metricValues.length;
+			}
+		}
+
+		return aggregatedMetrics;
+	}
+}

packages/cli/src/evaluation/test-runner/test-runner.service.ee.ts

@@ -15,11 +15,13 @@ import type { TestDefinition } from '@/databases/entities/test-definition.ee';
 import type { User } from '@/databases/entities/user';
 import type { WorkflowEntity } from '@/databases/entities/workflow-entity';
 import { ExecutionRepository } from '@/databases/repositories/execution.repository';
+import { TestMetricRepository } from '@/databases/repositories/test-metric.repository.ee';
 import { TestRunRepository } from '@/databases/repositories/test-run.repository.ee';
 import { WorkflowRepository } from '@/databases/repositories/workflow.repository';
 import { getRunData } from '@/workflow-execute-additional-data';
 import { WorkflowRunner } from '@/workflow-runner';
 
+import { EvaluationMetrics } from './evaluation-metrics.ee';
 import { createPinData, getPastExecutionStartNode } from './utils.ee';
 
 /**
@@ -40,6 +42,7 @@ export class TestRunnerService {
 		private readonly executionRepository: ExecutionRepository,
 		private readonly activeExecutions: ActiveExecutions,
 		private readonly testRunRepository: TestRunRepository,
+		private readonly testMetricRepository: TestMetricRepository,
 	) {}
 
 	/**
@@ -113,6 +116,11 @@ export class TestRunnerService {
 		return await executePromise;
 	}
 
+	/**
+	 * Evaluation result is the first item in the output of the last node
+	 * executed in the evaluation workflow. Defaults to an empty object
+	 * in case the node doesn't produce any output items.
+	 */
 	private extractEvaluationResult(execution: IRun): IDataObject {
 		const lastNodeExecuted = execution.data.resultData.lastNodeExecuted;
 		assert(lastNodeExecuted, 'Could not find the last node executed in evaluation workflow');
@@ -124,6 +132,21 @@ export class TestRunnerService {
 		return mainConnectionData?.[0]?.json ?? {};
 	}
 
+	/**
+	 * Get the metrics to collect from the evaluation workflow execution results.
+	 */
+	private async getTestMetricNames(testDefinitionId: string) {
+		const metrics = await this.testMetricRepository.find({
+			where: {
+				testDefinition: {
+					id: testDefinitionId,
+				},
+			},
+		});
+
+		return new Set(metrics.map((m) => m.name));
+	}
+
 	/**
 	 * Creates a new test run for the given test definition.
 	 */
@@ -152,11 +175,15 @@ export class TestRunnerService {
 				.andWhere('execution.workflowId = :workflowId', { workflowId: test.workflowId })
 				.getMany();
 
+		// Get the metrics to collect from the evaluation workflow
+		const testMetricNames = await this.getTestMetricNames(test.id);
+
 		// 2. Run over all the test cases
 
 		await this.testRunRepository.markAsRunning(testRun.id);
 
-		const metrics = [];
+		// Object to collect the results of the evaluation workflow executions
+		const metrics = new EvaluationMetrics(testMetricNames);
 
 		for (const { id: pastExecutionId } of pastExecutions) {
 			// Fetch past execution with data
@@ -192,12 +219,10 @@ export class TestRunnerService {
 			assert(evalExecution);
 
 			// Extract the output of the last node executed in the evaluation workflow
-			metrics.push(this.extractEvaluationResult(evalExecution));
+			metrics.addResults(this.extractEvaluationResult(evalExecution));
 		}
 
-		// TODO: 3. Aggregate the results
+		const aggregatedMetrics = metrics.getAggregatedMetrics();
-		// Now we just set success to true if all the test cases passed
-		const aggregatedMetrics = { success: metrics.every((metric) => metric.success) };
 
 		await this.testRunRepository.markAsCompleted(testRun.id, aggregatedMetrics);
 	}