@mintlify/scraping/src/root/retrieve.ts

Scrape documentation frameworks to Mintlify docs

import type { Root as HastRoot, Element } from 'hast';
import { visit, EXIT } from 'unist-util-visit';

import { framework } from '../utils/detectFramework.js';

export function retrieveRootContent(rootNode: HastRoot): Element | undefined {
  let rootTagName = 'main';
  switch (framework.vendor) {
    case 'docusaurus':
      rootTagName = 'article';
      break;
    case 'gitbook':
      rootTagName = 'main';
      break;
    case 'readme':
      rootTagName = 'article';
      break;
  }

  let rootSelector = 'break-anywhere';
  switch (framework.vendor) {
    case 'docusaurus':
      rootSelector = '';
      break;
    case 'gitbook':
      rootSelector = '';
      break;
    case 'readme':
      rootSelector = 'rm-Article';
      break;
  }

  let element: Element | undefined = undefined;
  visit(rootNode, 'element', function (node) {
    if (node.tagName === rootTagName) {
      if (rootSelector) {
        if (
          node.properties.className &&
          Array.isArray(node.properties.className) &&
          node.properties.className.includes(rootSelector)
        ) {
          element = node;
        }
      } else {
        element = node;
        return EXIT;
      }
    }
  });

  return element;
}