build-cdn-download-conf.ts 3.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596
  1. import path from 'node:path';
  2. import { fetchRemoteTextByLine, readFileIntoProcessedArray } from './lib/fetch-text-by-line';
  3. import { HostnameTrie } from './lib/trie';
  4. import { task } from './trace';
  5. import { SHARED_DESCRIPTION } from './constants/description';
  6. import { appendArrayInPlace } from './lib/append-array-in-place';
  7. import { SOURCE_DIR } from './constants/dir';
  8. import { DomainsetOutput } from './lib/rules/domainset';
  9. import { CRASHLYTICS_WHITELIST } from './constants/reject-data-source';
  10. import { appendSetElementsToArray } from 'foxts/append-set-elements-to-array';
  11. const getS3OSSDomainsPromise = (async (): Promise<Set<string>> => {
  12. const trie = new HostnameTrie();
  13. for await (const line of await fetchRemoteTextByLine('https://publicsuffix.org/list/public_suffix_list.dat', true)) {
  14. trie.add(line);
  15. }
  16. /**
  17. * Extract OSS domain from publicsuffix list
  18. */
  19. const S3OSSDomains = new Set<string>();
  20. trie.find('.amazonaws.com').forEach((line: string) => {
  21. if (
  22. (line.startsWith('s3-') || line.startsWith('s3.'))
  23. && !line.includes('cn-')
  24. ) {
  25. S3OSSDomains.add('.' + line);
  26. }
  27. });
  28. trie.find('.scw.cloud').forEach((line: string) => {
  29. if (
  30. (line.startsWith('s3-') || line.startsWith('s3.'))
  31. // && !line.includes('cn-')
  32. ) {
  33. S3OSSDomains.add('.' + line);
  34. }
  35. });
  36. trie.find('sakurastorage.jp').forEach((line: string) => {
  37. if (
  38. (line.startsWith('s3-') || line.startsWith('s3.'))
  39. ) {
  40. S3OSSDomains.add('.' + line);
  41. }
  42. });
  43. return S3OSSDomains;
  44. })();
  45. const cdnDomainsListPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'domainset/cdn.conf'));
  46. const downloadDomainSetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'domainset/download.conf'));
  47. const steamDomainSetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'domainset/game-download.conf'));
  48. export const buildCdnDownloadConf = task(require.main === module, __filename)(async (span) => {
  49. const [
  50. S3OSSDomains,
  51. cdnDomainsList,
  52. downloadDomainSet,
  53. steamDomainSet
  54. ] = await Promise.all([
  55. span.traceChildPromise('download public suffix list for s3', getS3OSSDomainsPromise),
  56. cdnDomainsListPromise,
  57. downloadDomainSetPromise,
  58. steamDomainSetPromise
  59. ]);
  60. // Move S3 domains to download domain set, since S3 files may be large
  61. appendSetElementsToArray(downloadDomainSet, S3OSSDomains);
  62. appendArrayInPlace(downloadDomainSet, steamDomainSet);
  63. // we have whitelisted the crashlytics domain, and we also want to put it in CDN policy
  64. appendArrayInPlace(cdnDomainsList, CRASHLYTICS_WHITELIST);
  65. return Promise.all([
  66. new DomainsetOutput(span, 'cdn')
  67. .withTitle('Sukka\'s Ruleset - CDN Domains')
  68. .withDescription([
  69. ...SHARED_DESCRIPTION,
  70. '',
  71. 'This file contains object storage and static assets CDN domains.'
  72. ])
  73. .addFromDomainset(cdnDomainsList)
  74. .write(),
  75. new DomainsetOutput(span, 'download')
  76. .withTitle('Sukka\'s Ruleset - Large Files Hosting Domains')
  77. .withDescription([
  78. ...SHARED_DESCRIPTION,
  79. '',
  80. 'This file contains domains for software updating & large file hosting.'
  81. ])
  82. .addFromDomainset(downloadDomainSet)
  83. .write()
  84. ]);
  85. });