nord-for-org/web/dart/parse_sitemap.dart

import 'dart:html' as html show HttpRequest, Element, querySelector;

import 'package:html/parser.dart' show parse;
import 'package:html/dom.dart' as dom show Element;

final excluded_keywords = {'index', 'CONTRIBUTING', 'LICENSE', 'README'};

// Get the sitemap content
Future<String> fetchRemoteSitemap() async {
  const path = 'sitemap.html';
  try {
    return await html.HttpRequest.getString(path);
  } catch (e) {
    print('Couldn’t open $path');
  }
  return 'Error';
}

// Parse the list of elements and detect pages from this list
Map<String, String> detectPages(List<dom.Element> t_sitemap,
    [String t_prefix]) {
  final sitemap = t_sitemap.where((e) => !excluded_keywords.contains(e));
  final links = <String, String>{};
  for (var elem in sitemap) {
    if (elem.innerHtml.startsWith('<a')) {
      elem = elem.firstChild;
      final url = elem.attributes['href'];
      final text = elem.firstChild.text;
      links[url] = (t_prefix == null) ? text : '$text ($t_prefix)';
    } else {
      t_prefix = (t_prefix == null)
          ? elem.firstChild.text.replaceAll('\n', '')
          : '$t_prefix / ${elem.firstChild.text}';
      final ul = elem.children[0].children;
      links.addAll(detectPages(ul, t_prefix));
    }
  }
  return links;
}

// This function returns a Map which contains all links to languages detected
// from the sitemap.
Future<Map<String, String>> parseSitemap() async {
  final content = await fetchRemoteSitemap();
  final sitemap = parse(content).getElementsByClassName('org-ul')[0].children;
  return detectPages(sitemap);
}

Future<void> createSitemap() async {
  final sitemap = await parseSitemap();
  final pages = html.querySelector('#drop-page');
  sitemap.forEach((url, name) {
    final link = html.Element.li()
      ..classes.add('dropdown-item')
      ..append(html.Element.a()
        ..attributes['href'] = url
        ..innerText = name);
    pages.append(link);
  });
}