feat(Structured Output Parser Node): Refactor Output Parsers and Improve Error Handling (#11148)

This commit is contained in:
oleg
2024-10-22 10:46:58 +02:00
committed by GitHub
parent 4dde772814
commit 45274f2e7f
20 changed files with 1061 additions and 383 deletions

View File

@@ -1,8 +1,13 @@
import type { IExecuteFunctions, INode, IWorkflowDataProxyData } from 'n8n-workflow';
import { mock } from 'jest-mock-extended';
import { normalizeItems } from 'n8n-core';
import type { z } from 'zod';
import type { StructuredOutputParser } from 'langchain/output_parsers';
import {
jsonParse,
type IExecuteFunctions,
type INode,
type IWorkflowDataProxyData,
} from 'n8n-workflow';
import type { N8nStructuredOutputParser } from '../../../../utils/output_parsers/N8nStructuredOutputParser';
import { OutputParserStructured } from '../OutputParserStructured.node';
describe('OutputParserStructured', () => {
@@ -11,139 +16,451 @@ describe('OutputParserStructured', () => {
helpers: { normalizeItems },
});
const workflowDataProxy = mock<IWorkflowDataProxyData>({ $input: mock() });
thisArg.getWorkflowDataProxy.mockReturnValue(workflowDataProxy);
thisArg.getNode.mockReturnValue(mock<INode>({ typeVersion: 1.1 }));
thisArg.addInputData.mockReturnValue({ index: 0 });
thisArg.addOutputData.mockReturnValue();
beforeEach(() => {
outputParser = new OutputParserStructured();
thisArg.getWorkflowDataProxy.mockReturnValue(workflowDataProxy);
thisArg.addInputData.mockReturnValue({ index: 0 });
thisArg.addOutputData.mockReturnValue();
});
describe('supplyData', () => {
it('should parse a valid JSON schema', async () => {
const schema = `{
"type": "object",
"properties": {
"name": {
"type": "string"
describe('Version 1.1 and below', () => {
beforeEach(() => {
thisArg.getNode.mockReturnValue(mock<INode>({ typeVersion: 1.1 }));
});
it('should parse a complex nested schema', async () => {
const schema = `{
"type": "object",
"properties": {
"user": {
"type": "object",
"properties": {
"name": { "type": "string" },
"details": {
"type": "object",
"properties": {
"age": { "type": "number" },
"hobbies": { "type": "array", "items": { "type": "string" } }
}
}
}
},
"timestamp": { "type": "string", "format": "date-time" }
},
"age": {
"type": "number"
}
},
"required": ["name", "age"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: StructuredOutputParser<z.ZodType<object, z.ZodTypeDef, object>>;
};
const outputObject = { output: { name: 'Mac', age: 27 } };
const parsersOutput = await response.parse(`Here's the output!
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
});
it('should handle missing required properties', async () => {
const schema = `{
"type": "object",
"properties": {
"name": {
"type": "string"
},
"age": {
"type": "number"
}
},
"required": ["name", "age"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: StructuredOutputParser<z.ZodType<object, z.ZodTypeDef, object>>;
};
const outputObject = { output: { name: 'Mac' } };
await expect(
response.parse(`Here's the output!
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`),
).rejects.toThrow('Required');
});
it('should throw on wrong type', async () => {
const schema = `{
"type": "object",
"properties": {
"name": {
"type": "string"
},
"age": {
"type": "number"
}
},
"required": ["name", "age"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: StructuredOutputParser<z.ZodType<object, z.ZodTypeDef, object>>;
};
const outputObject = { output: { name: 'Mac', age: '27' } };
await expect(
response.parse(`Here's the output!
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`),
).rejects.toThrow('Expected number, received string');
});
it('should parse array output', async () => {
const schema = `{
"type": "object",
"properties": {
"myArr": {
"type": "array",
"items": {
"type": "object",
"properties": {
"name": {
"type": "string"
},
"age": {
"type": "number"
}
"required": ["user", "timestamp"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: {
user: {
name: 'Alice',
details: {
age: 30,
hobbies: ['reading', 'hiking'],
},
"required": ["name", "age"]
}
}
},
"required": ["myArr"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: StructuredOutputParser<z.ZodType<object, z.ZodTypeDef, object>>;
};
const outputObject = {
output: {
myArr: [
{ name: 'Mac', age: 27 },
{ name: 'Alice', age: 25 },
],
},
};
const parsersOutput = await response.parse(`Here's the output!
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
},
timestamp: '2023-04-01T12:00:00Z',
},
};
const parsersOutput = await response.parse(`Here's the complex output:
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
expect(parsersOutput).toEqual(outputObject);
});
it('should handle optional fields correctly', async () => {
const schema = `{
"type": "object",
"properties": {
"name": { "type": "string" },
"age": { "type": "number" },
"email": { "type": "string", "format": "email" }
},
"required": ["name"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: {
name: 'Bob',
email: 'bob@example.com',
},
};
const parsersOutput = await response.parse(`Here's the output with optional fields:
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
});
it('should handle arrays of objects', async () => {
const schema = `{
"type": "object",
"properties": {
"users": {
"type": "array",
"items": {
"type": "object",
"properties": {
"id": { "type": "number" },
"name": { "type": "string" }
},
"required": ["id", "name"]
}
}
},
"required": ["users"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: {
users: [
{ id: 1, name: 'Alice' },
{ id: 2, name: 'Bob' },
],
},
};
const parsersOutput = await response.parse(`Here's the array output:
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
});
it('should handle empty objects', async () => {
const schema = `{
"type": "object",
"properties": {
"data": {
"type": "object"
}
},
"required": ["data"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: {
data: {},
},
};
const parsersOutput = await response.parse(`Here's the empty object output:
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
});
it('should throw error for null values in non-nullable fields', async () => {
const schema = `{
"type": "object",
"properties": {
"name": { "type": "string" },
"age": { "type": "number" }
},
"required": ["name", "age"]
}`;
thisArg.getNodeParameter.calledWith('jsonSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: {
name: 'Charlie',
age: null,
},
};
await expect(
response.parse(
`Here's the output with null value:
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`,
undefined,
(e) => e,
),
).rejects.toThrow('Expected number, received null');
});
});
describe('Version 1.2 and above', () => {
beforeEach(() => {
thisArg.getNode.mockReturnValue(mock<INode>({ typeVersion: 1.2 }));
});
it('should parse output using schema generated from complex JSON example', async () => {
const jsonExample = `{
"user": {
"name": "Alice",
"details": {
"age": 30,
"address": {
"street": "123 Main St",
"city": "Anytown",
"zipCode": "12345"
}
}
},
"orders": [
{
"id": "ORD-001",
"items": ["item1", "item2"],
"total": 50.99
},
{
"id": "ORD-002",
"items": ["item3"],
"total": 25.50
}
],
"isActive": true
}`;
thisArg.getNodeParameter.calledWith('schemaType', 0).mockReturnValueOnce('fromJson');
thisArg.getNodeParameter
.calledWith('jsonSchemaExample', 0)
.mockReturnValueOnce(jsonExample);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: jsonParse(jsonExample),
};
const parsersOutput = await response.parse(`Here's the complex output:
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
});
it('should validate enum values', async () => {
const inputSchema = `{
"type": "object",
"properties": {
"color": {
"type": "string",
"enum": ["red", "green", "blue"]
}
},
"required": ["color"]
}`;
thisArg.getNodeParameter.calledWith('schemaType', 0).mockReturnValueOnce('manual');
thisArg.getNodeParameter.calledWith('inputSchema', 0).mockReturnValueOnce(inputSchema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const validOutput = {
output: {
color: 'green',
},
};
const invalidOutput = {
output: {
color: 'yellow',
},
};
await expect(
response.parse(`Valid output:
\`\`\`json
${JSON.stringify(validOutput)}
\`\`\`
`),
).resolves.toEqual(validOutput);
await expect(
response.parse(
`Invalid output:
\`\`\`json
${JSON.stringify(invalidOutput)}
\`\`\`
`,
undefined,
(e) => e,
),
).rejects.toThrow();
});
it('should handle recursive structures', async () => {
const inputSchema = `{
"type": "object",
"properties": {
"name": { "type": "string" },
"children": {
"type": "array",
"items": { "$ref": "#" }
}
},
"required": ["name"]
}`;
thisArg.getNodeParameter.calledWith('schemaType', 0).mockReturnValueOnce('manual');
thisArg.getNodeParameter.calledWith('inputSchema', 0).mockReturnValueOnce(inputSchema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: {
name: 'Root',
children: [
{
name: 'Child1',
children: [{ name: 'Grandchild1' }, { name: 'Grandchild2' }],
},
{
name: 'Child2',
},
],
},
};
const parsersOutput = await response.parse(`Here's the recursive structure output:
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
});
it('should handle missing required properties', async () => {
const schema = `{
"type": "object",
"properties": {
"name": {
"type": "string"
},
"age": {
"type": "number"
}
},
"required": ["name", "age"]
}`;
thisArg.getNodeParameter.calledWith('schemaType', 0).mockReturnValueOnce('manual');
thisArg.getNodeParameter.calledWith('inputSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = { output: { name: 'Mac' } };
await expect(
response.parse(
`Here's the output!
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`,
undefined,
(e) => e,
),
).rejects.toThrow('Required');
});
it('should throw on wrong type', async () => {
const schema = `{
"type": "object",
"properties": {
"name": {
"type": "string"
},
"age": {
"type": "number"
}
},
"required": ["name", "age"]
}`;
thisArg.getNodeParameter.calledWith('inputSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = { output: { name: 'Mac', age: '27' } };
await expect(
response.parse(
`Here's the output!
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`,
undefined,
(e) => e,
),
).rejects.toThrow('Expected number, received string');
});
it('should parse array output', async () => {
const schema = `{
"type": "object",
"properties": {
"myArr": {
"type": "array",
"items": {
"type": "object",
"properties": {
"name": {
"type": "string"
},
"age": {
"type": "number"
}
},
"required": ["name", "age"]
}
}
},
"required": ["myArr"]
}`;
thisArg.getNodeParameter.calledWith('inputSchema', 0).mockReturnValueOnce(schema);
const { response } = (await outputParser.supplyData.call(thisArg, 0)) as {
response: N8nStructuredOutputParser;
};
const outputObject = {
output: {
myArr: [
{ name: 'Mac', age: 27 },
{ name: 'Alice', age: 25 },
],
},
};
const parsersOutput = await response.parse(`Here's the output!
\`\`\`json
${JSON.stringify(outputObject)}
\`\`\`
`);
expect(parsersOutput).toEqual(outputObject);
});
});
});
});