LangChain Stripe Loader

A LangChain document loader for Stripe documentation and website content. This package provides tools to extract and process content from Stripe's documentation and website for use with LangChain applications.

Installation

npm install langchain-stripe-loader

Features

• Extract content from Stripe's official documentation (docs.stripe.com)
• Extract content from Stripe's main website (stripe.com)
• Support for different locales
• Convert HTML content to Markdown for better compatibility with LLMs
• Metadata extraction (title, description, source URL)

Usage

Loading Content from Stripe Documentation

import { StripeDocsDocumentLoader } from 'langchain-stripe-loader';

// Create a loader instance
const loader = new StripeDocsDocumentLoader();

// Load documents with default locale (en-US)
const documents = await loader.load();

// Or specify a different locale
const jaDocuments = await loader.load({
  locale: 'ja-JP',
});

// Load documents with specific options
const customDocuments = await loader.load({
  resource: 'connect', // Filter by resource path
  excludeResources: ['connect/account-balances'],
});

// Use the documents with your LangChain application
console.log(`Loaded ${documents.length} documents`);

Loading Content from Stripe.com Website

import { StripeComDocumentLoader } from 'langchain-stripe-loader';

// Create a loader instance
const loader = new StripeComDocumentLoader();

// Load all documents
const documents = await loader.load();

// Load documents with specific options
const customDocuments = await loader.load({
  resource: '/jobs', // Filter by resource path
  locale: 'en-US', // Specify locale
});

// Load specific URLs
const specificDocuments = await loader.load({
  urls: ['https://stripe.com/docs/payments', 'https://stripe.com/docs/billing'],
});

API Reference

StripeDocsDocumentLoader

Document loader for Stripe's official documentation (docs.stripe.com).

Constructor

constructor(debug: boolean = false)

• debug: Enable debug logging

Methods

• load(locale: string = 'en-US'): Promise<Document[]> - Load documents from Stripe Docs with the specified locale

StripeComDocumentLoader

Document loader for Stripe's main website (stripe.com).

Constructor

constructor(debug: boolean = false)

• debug: Enable debug logging

Methods

• load(options?: { resource?: string; locale?: string; urls?: string[] }): Promise<Document[]> - Load documents from Stripe.com with optional filtering

How It Works

1. The loaders fetch URLs from Stripe's sitemaps
2. HTML content is downloaded from each URL
3. Relevant content is extracted using regex patterns
4. HTML is converted to Markdown using Turndown
5. Content is wrapped in LangChain Document objects with metadata

Dependencies

• @langchain/core: LangChain core functionality
• stripe-loaders-core: Core utilities for Stripe loaders
• turndown: HTML to Markdown conversion

License

MIT