feat(OpenAI Node): Add support for ChatGPT (#5596)

packages/nodes-base/nodes/OpenAi/ChatDescription.ts

@@ -0,0 +1,313 @@
+import type { INodeExecutionData, INodeProperties } from 'n8n-workflow';
+
+export const chatOperations: INodeProperties[] = [
+	{
+		displayName: 'Operation',
+		name: 'operation',
+		type: 'options',
+		noDataExpression: true,
+		displayOptions: {
+			show: {
+				resource: ['chat'],
+			},
+		},
+		options: [
+			{
+				name: 'Complete',
+				value: 'complete',
+				action: 'Create a Completion',
+				description: 'Create one or more completions for a given text',
+				routing: {
+					request: {
+						method: 'POST',
+						url: '/v1/chat/completions',
+					},
+				},
+			},
+		],
+		default: 'complete',
+	},
+];
+
+const completeOperations: INodeProperties[] = [
+	{
+		displayName: 'Model',
+		name: 'model',
+		type: 'options',
+		description:
+			'The model to use. Currently, only gpt-3.5-turbo and gpt-3.5-turbo-0301 are supported.',
+		displayOptions: {
+			show: {
+				operation: ['complete'],
+				resource: ['chat'],
+			},
+		},
+		options: [
+			{
+				name: 'gpt-3.5-turbo',
+				value: 'gpt-3.5-turbo',
+			},
+			{
+				name: 'gpt-3.5-turbo-0301',
+				value: 'gpt-3.5-turbo-0301',
+			},
+		],
+		routing: {
+			send: {
+				type: 'body',
+				property: 'model',
+			},
+		},
+		default: 'gpt-3.5-turbo',
+	},
+	{
+		displayName: 'Prompt',
+		name: 'prompt',
+		type: 'fixedCollection',
+		typeOptions: {
+			sortable: true,
+			multipleValues: true,
+		},
+		displayOptions: {
+			show: {
+				resource: ['chat'],
+				operation: ['complete'],
+			},
+		},
+		placeholder: 'Add Message',
+		default: {},
+		options: [
+			{
+				displayName: 'Messages',
+				name: 'messages',
+				values: [
+					{
+						displayName: 'Role',
+						name: 'role',
+						type: 'options',
+						options: [
+							{
+								name: 'Assistant',
+								value: 'assistant',
+							},
+							{
+								name: 'System',
+								value: 'system',
+							},
+							{
+								name: 'User',
+								value: 'user',
+							},
+						],
+						default: 'user',
+					},
+					{
+						displayName: 'Content',
+						name: 'content',
+						type: 'string',
+						default: '',
+					},
+				],
+			},
+		],
+		routing: {
+			send: {
+				type: 'body',
+				property: 'messages',
+				value: '={{ $parameter.prompt.messages }}',
+			},
+		},
+	},
+];
+
+const sharedOperations: INodeProperties[] = [
+	{
+		displayName: 'Simplify',
+		name: 'simplifyOutput',
+		type: 'boolean',
+		default: true,
+		displayOptions: {
+			show: {
+				operation: ['complete'],
+				resource: ['chat'],
+			},
+		},
+		routing: {
+			output: {
+				postReceive: [
+					{
+						type: 'set',
+						enabled: '={{$value}}',
+						properties: {
+							value: '={{ { "data": $response.body.choices } }}',
+						},
+					},
+					{
+						type: 'rootProperty',
+						enabled: '={{$value}}',
+						properties: {
+							property: 'data',
+						},
+					},
+					async function (items: INodeExecutionData[]): Promise<INodeExecutionData[]> {
+						if (this.getNode().parameters.simplifyOutput === false) {
+							return items;
+						}
+						return items.map((item) => {
+							return {
+								json: {
+									...item.json,
+									message: item.json.message,
+								},
+							};
+						});
+					},
+				],
+			},
+		},
+		description: 'Whether to return a simplified version of the response instead of the raw data',
+	},
+
+	{
+		displayName: 'Options',
+		name: 'options',
+		placeholder: 'Add Option',
+		description: 'Additional options to add',
+		type: 'collection',
+		default: {},
+		displayOptions: {
+			show: {
+				operation: ['complete'],
+				resource: ['chat'],
+			},
+		},
+		options: [
+			{
+				displayName: 'Echo Prompt',
+				name: 'echo',
+				type: 'boolean',
+				description: 'Whether the prompt should be echo back in addition to the completion',
+				default: false,
+				displayOptions: {
+					show: {
+						'/operation': ['complete'],
+					},
+				},
+				routing: {
+					send: {
+						type: 'body',
+						property: 'echo',
+					},
+				},
+			},
+			{
+				displayName: 'Frequency Penalty',
+				name: 'frequency_penalty',
+				default: 0,
+				typeOptions: { maxValue: 2, minValue: -2, numberPrecision: 1 },
+				description:
+					"Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim",
+				type: 'number',
+				routing: {
+					send: {
+						type: 'body',
+						property: 'frequency_penalty',
+					},
+				},
+			},
+			{
+				displayName: 'Maximum Number of Tokens',
+				name: 'maxTokens',
+				default: 16,
+				description:
+					'The maximum number of tokens to generate in the completion. Most models have a context length of 2048 tokens (except for the newest models, which support 4096).',
+				type: 'number',
+				displayOptions: {
+					show: {
+						'/operation': ['complete'],
+					},
+				},
+				typeOptions: {
+					maxValue: 4096,
+				},
+				routing: {
+					send: {
+						type: 'body',
+						property: 'max_tokens',
+					},
+				},
+			},
+			{
+				displayName: 'Number of Completions',
+				name: 'n',
+				default: 1,
+				description:
+					'How many completions to generate for each prompt. Note: Because this parameter generates many completions, it can quickly consume your token quota. Use carefully and ensure that you have reasonable settings for max_tokens and stop.',
+				type: 'number',
+				routing: {
+					send: {
+						type: 'body',
+						property: 'n',
+					},
+				},
+			},
+			{
+				displayName: 'Presence Penalty',
+				name: 'presence_penalty',
+				default: 0,
+				typeOptions: { maxValue: 2, minValue: -2, numberPrecision: 1 },
+				description:
+					"Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics",
+				type: 'number',
+				routing: {
+					send: {
+						type: 'body',
+						property: 'presence_penalty',
+					},
+				},
+			},
+			{
+				displayName: 'Sampling Temperature',
+				name: 'temperature',
+				default: 1,
+				typeOptions: { maxValue: 1, minValue: 0, numberPrecision: 1 },
+				description:
+					'Controls randomness: Lowering results in less random completions. As the temperature approaches zero, the model will become deterministic and repetitive.',
+				type: 'number',
+				routing: {
+					send: {
+						type: 'body',
+						property: 'temperature',
+					},
+				},
+			},
+			{
+				displayName: 'Top P',
+				name: 'topP',
+				default: 1,
+				typeOptions: { maxValue: 1, minValue: 0, numberPrecision: 1 },
+				description:
+					'Controls diversity via nucleus sampling: 0.5 means half of all likelihood-weighted options are considered. We generally recommend altering this or temperature but not both.',
+				type: 'number',
+				routing: {
+					send: {
+						type: 'body',
+						property: 'top_p',
+					},
+				},
+			},
+		],
+	},
+];
+
+export const chatFields: INodeProperties[] = [
+	/* -------------------------------------------------------------------------- */
+	/*                               chat:complete                        */
+	/* -------------------------------------------------------------------------- */
+	...completeOperations,
+
+	/* -------------------------------------------------------------------------- */
+	/*                                chat:ALL                                    */
+	/* -------------------------------------------------------------------------- */
+	...sharedOperations,
+];

packages/nodes-base/nodes/OpenAi/OpenAi.node.ts

@@ -1,6 +1,7 @@
 import type { INodeType, INodeTypeDescription } from 'n8n-workflow';
 import { imageFields, imageOperations } from './ImageDescription';
 import { textFields, textOperations } from './TextDescription';
+import { chatFields, chatOperations } from './ChatDescription';
 
 export class OpenAi implements INodeType {
 	description: INodeTypeDescription = {
@@ -32,6 +33,10 @@ export class OpenAi implements INodeType {
 				type: 'options',
 				noDataExpression: true,
 				options: [
+					{
+						name: 'Chat',
+						value: 'chat',
+					},
 					{
 						name: 'Image',
 						value: 'image',
@@ -44,6 +49,9 @@ export class OpenAi implements INodeType {
 				default: 'text',
 			},
 
+			...chatOperations,
+			...chatFields,
+
 			...imageOperations,
 			...imageFields,