first commit

packages/nodes-base/nodes/Airtop/actions/extraction/Extraction.resource.ts

@@ -0,0 +1,46 @@
+import type { INodeProperties } from 'n8n-workflow';
+
+import * as getPaginated from './getPaginated.operation';
+import * as query from './query.operation';
+import * as scrape from './scrape.operation';
+import { getSessionModeFields } from '../common/fields';
+
+export { getPaginated, query, scrape };
+
+export const description: INodeProperties[] = [
+	{
+		displayName: 'Operation',
+		name: 'operation',
+		type: 'options',
+		noDataExpression: true,
+		displayOptions: {
+			show: {
+				resource: ['extraction'],
+			},
+		},
+		options: [
+			{
+				name: 'Query Page',
+				value: 'query',
+				description: 'Query a page to extract data or ask a question given the data on the page',
+				action: 'Query page',
+			},
+			{
+				name: 'Query Page with Pagination',
+				value: 'getPaginated',
+				description: 'Extract content from paginated or dynamically loaded pages',
+				action: 'Query page with pagination',
+			},
+			{
+				name: 'Smart Scrape',
+				value: 'scrape',
+				description: 'Scrape a page and return the data as markdown',
+				action: 'Smart scrape page',
+			},
+		],
+		default: 'getPaginated',
+	},
+	...getSessionModeFields('extraction', ['getPaginated', 'query', 'scrape']),
+	...getPaginated.description,
+	...query.description,
+];

packages/nodes-base/nodes/Airtop/actions/extraction/getPaginated.operation.ts

@@ -0,0 +1,125 @@
+import {
+	type IExecuteFunctions,
+	type INodeExecutionData,
+	type INodeProperties,
+} from 'n8n-workflow';
+
+import { outputSchemaField, parseJsonOutputField } from '../common/fields';
+import { parseJsonIfPresent } from '../common/output.utils';
+import { executeRequestWithSessionManagement } from '../common/session.utils';
+
+export const description: INodeProperties[] = [
+	{
+		displayName: 'Prompt',
+		name: 'prompt',
+		type: 'string',
+		typeOptions: {
+			rows: 4,
+		},
+		required: true,
+		default: '',
+		displayOptions: {
+			show: {
+				resource: ['extraction'],
+				operation: ['getPaginated'],
+			},
+		},
+		description: 'The prompt to extract data from the pages',
+		placeholder: 'e.g. Extract all the product names and prices',
+	},
+	{
+		displayName: 'Additional Fields',
+		name: 'additionalFields',
+		type: 'collection',
+		placeholder: 'Add Field',
+		default: {},
+		displayOptions: {
+			show: {
+				resource: ['extraction'],
+				operation: ['getPaginated'],
+			},
+		},
+		options: [
+			{
+				...outputSchemaField,
+			},
+			{
+				...parseJsonOutputField,
+			},
+			{
+				displayName: 'Interaction Mode',
+				name: 'interactionMode',
+				type: 'options',
+				default: 'auto',
+				description: 'The strategy for interacting with the page',
+				options: [
+					{
+						name: 'Auto',
+						description: 'Automatically choose the most cost-effective mode',
+						value: 'auto',
+					},
+					{
+						name: 'Accurate',
+						description: 'Prioritize accuracy over cost',
+						value: 'accurate',
+					},
+					{
+						name: 'Cost Efficient',
+						description: 'Minimize costs while ensuring effectiveness',
+						value: 'cost-efficient',
+					},
+				],
+			},
+			{
+				displayName: 'Pagination Mode',
+				name: 'paginationMode',
+				type: 'options',
+				default: 'auto',
+				description: 'The pagination approach to use',
+				options: [
+					{
+						name: 'Auto',
+						description: 'Look for pagination links first, then try infinite scrolling',
+						value: 'auto',
+					},
+					{
+						name: 'Paginated',
+						description: 'Only use pagination links',
+						value: 'paginated',
+					},
+					{
+						name: 'Infinite Scroll',
+						description: 'Scroll the page to load more content',
+						value: 'infinite-scroll',
+					},
+				],
+			},
+		],
+	},
+];
+
+export async function execute(
+	this: IExecuteFunctions,
+	index: number,
+): Promise<INodeExecutionData[]> {
+	const prompt = this.getNodeParameter('prompt', index, '') as string;
+	const additionalFields = this.getNodeParameter('additionalFields', index);
+
+	const configFields = ['paginationMode', 'interactionMode', 'outputSchema'];
+	const configuration = configFields.reduce(
+		(config, key) => (additionalFields[key] ? { ...config, [key]: additionalFields[key] } : config),
+		{},
+	);
+
+	const result = await executeRequestWithSessionManagement.call(this, index, {
+		method: 'POST',
+		path: '/sessions/{sessionId}/windows/{windowId}/paginated-extraction',
+		body: {
+			prompt,
+			configuration,
+		},
+	});
+
+	const nodeOutput = parseJsonIfPresent.call(this, index, result);
+	return this.helpers.returnJsonArray(nodeOutput);
+}

packages/nodes-base/nodes/Airtop/actions/extraction/query.operation.ts

@@ -0,0 +1,85 @@
+import {
+	type IExecuteFunctions,
+	type INodeExecutionData,
+	type INodeProperties,
+} from 'n8n-workflow';
+
+import { outputSchemaField, parseJsonOutputField } from '../common/fields';
+import { parseJsonIfPresent } from '../common/output.utils';
+import { executeRequestWithSessionManagement } from '../common/session.utils';
+
+export const description: INodeProperties[] = [
+	{
+		displayName: 'Prompt',
+		name: 'prompt',
+		type: 'string',
+		typeOptions: {
+			rows: 4,
+		},
+		required: true,
+		default: '',
+		placeholder: 'e.g. Is there a login form in this page?',
+		displayOptions: {
+			show: {
+				resource: ['extraction'],
+				operation: ['query'],
+			},
+		},
+		description: 'The prompt to query the page content',
+	},
+	{
+		displayName: 'Additional Fields',
+		name: 'additionalFields',
+		type: 'collection',
+		placeholder: 'Add Field',
+		default: {},
+		displayOptions: {
+			show: {
+				resource: ['extraction'],
+				operation: ['query'],
+			},
+		},
+		options: [
+			{
+				...outputSchemaField,
+			},
+			{
+				...parseJsonOutputField,
+			},
+			{
+				displayName: 'Include Visual Analysis',
+				name: 'includeVisualAnalysis',
+				type: 'boolean',
+				default: false,
+				description: 'Whether to analyze the web page visually when fulfilling the request',
+			},
+		],
+	},
+];
+
+export async function execute(
+	this: IExecuteFunctions,
+	index: number,
+): Promise<INodeExecutionData[]> {
+	const prompt = this.getNodeParameter('prompt', index, '') as string;
+	const additionalFields = this.getNodeParameter('additionalFields', index, {});
+	const outputSchema = additionalFields.outputSchema;
+	const includeVisualAnalysis = additionalFields.includeVisualAnalysis;
+
+	const result = await executeRequestWithSessionManagement.call(this, index, {
+		method: 'POST',
+		path: '/sessions/{sessionId}/windows/{windowId}/page-query',
+		body: {
+			prompt,
+			configuration: {
+				experimental: {
+					includeVisualAnalysis: includeVisualAnalysis ? 'enabled' : 'disabled',
+				},
+				...(outputSchema ? { outputSchema } : {}),
+			},
+		},
+	});
+
+	const nodeOutput = parseJsonIfPresent.call(this, index, result);
+	return this.helpers.returnJsonArray(nodeOutput);
+}

packages/nodes-base/nodes/Airtop/actions/extraction/scrape.operation.ts

@@ -0,0 +1,16 @@
+import type { IExecuteFunctions, INodeExecutionData } from 'n8n-workflow';
+
+import { executeRequestWithSessionManagement } from '../common/session.utils';
+
+export async function execute(
+	this: IExecuteFunctions,
+	index: number,
+): Promise<INodeExecutionData[]> {
+	const result = await executeRequestWithSessionManagement.call(this, index, {
+		method: 'POST',
+		path: '/sessions/{sessionId}/windows/{windowId}/scrape-content',
+		body: {},
+	});
+
+	return this.helpers.returnJsonArray({ ...result });
+}