build-microsoft-cdn.ts 2.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869
  1. import { task } from './trace';
  2. import { fetchRemoteTextByLine } from './lib/fetch-text-by-line';
  3. import { HostnameSmolTrie } from './lib/trie';
  4. import { SHARED_DESCRIPTION } from './lib/constants';
  5. import { createMemoizedPromise } from './lib/memo-promise';
  6. import { extractDomainsFromFelixDnsmasq } from './lib/parse-dnsmasq';
  7. import { RulesetOutput } from './lib/create-file';
  8. import { appendArrayInPlace } from './lib/append-array-in-place';
  9. const PROBE_DOMAINS = ['.microsoft.com', '.windows.net', '.windows.com', '.windowsupdate.com', '.windowssearch.com', '.office.net'];
  10. const DOMAINS = [
  11. 'res.cdn.office.net',
  12. 'res-1.cdn.office.net',
  13. 'statics.teams.cdn.office.net'
  14. ];
  15. const DOMAIN_SUFFIXES = ['download.prss.microsoft.com'];
  16. const BLACKLIST = [
  17. 'www.microsoft.com',
  18. 'learn.microsoft.com',
  19. 'devblogs.microsoft.com',
  20. 'docs.microsoft.com',
  21. 'developer.microsoft.com',
  22. 'windowsupdate.com'
  23. ];
  24. export const getMicrosoftCdnRulesetPromise = createMemoizedPromise<[domains: string[], domainSuffixes: string[]]>(async () => {
  25. // First trie is to find the microsoft domains that matches probe domains
  26. const trie = new HostnameSmolTrie();
  27. for await (const line of await fetchRemoteTextByLine('https://raw.githubusercontent.com/felixonmars/dnsmasq-china-list/master/accelerated-domains.china.conf')) {
  28. const domain = extractDomainsFromFelixDnsmasq(line);
  29. if (domain) {
  30. trie.add(domain);
  31. }
  32. }
  33. const foundMicrosoftCdnDomains = PROBE_DOMAINS.flatMap(domain => trie.find(domain));
  34. // Second trie is to remove blacklisted domains
  35. const trie2 = new HostnameSmolTrie(foundMicrosoftCdnDomains);
  36. BLACKLIST.forEach(black => trie2.whitelist(black));
  37. const domains: string[] = DOMAINS;
  38. const domainSuffixes: string[] = DOMAIN_SUFFIXES;
  39. appendArrayInPlace(domainSuffixes, trie2.dump());
  40. return [domains, domainSuffixes] as const;
  41. });
  42. export const buildMicrosoftCdn = task(require.main === module, __filename)(async (span) => {
  43. const description = [
  44. ...SHARED_DESCRIPTION,
  45. '',
  46. 'This file contains Microsoft\'s domains using their China mainland CDN servers.',
  47. '',
  48. 'Data from:',
  49. ' - https://github.com/felixonmars/dnsmasq-china-list'
  50. ];
  51. const [domains, domainSuffixes] = await span.traceChildPromise('get microsoft cdn domains', getMicrosoftCdnRulesetPromise());
  52. return new RulesetOutput(span, 'microsoft_cdn', 'non_ip')
  53. .withTitle('Sukka\'s Ruleset - Microsoft CDN')
  54. .withDescription(description)
  55. .bulkAddDomain(domains)
  56. .bulkAddDomainSuffix(domainSuffixes)
  57. .write();
  58. });