build-internal-cdn-rules.js 3.5 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798
  1. // @ts-check
  2. const fsp = require('fs/promises');
  3. const path = require('path');
  4. const tldts = require('tldts');
  5. const { processLine } = require('./lib/process-line');
  6. const { readFileByLine } = require('./lib/fetch-remote-text-by-line');
  7. const { createDomainSorter } = require('./lib/stable-sort-domain');
  8. const { task } = require('./lib/trace-runner');
  9. const { compareAndWriteFile } = require('./lib/create-file');
  10. const { getGorhillPublicSuffixPromise } = require('./lib/get-gorhill-publicsuffix');
  11. // const { createCachedGorhillGetDomain } = require('./lib/cached-tld-parse');
  12. const escapeRegExp = (string = '') => string.replaceAll(/[$()*+.?[\\\]^{|}]/g, '\\$&');
  13. const buildInternalCDNDomains = task(__filename, async () => {
  14. const set = new Set();
  15. const keywords = new Set();
  16. const gorhill = await getGorhillPublicSuffixPromise();
  17. const domainSorter = createDomainSorter(gorhill);
  18. /**
  19. * @param {string} input
  20. */
  21. const addApexDomain = (input) => {
  22. // We are including the private domains themselves
  23. const d = tldts.getDomain(input, { allowPrivateDomains: false });
  24. if (d) {
  25. set.add(d);
  26. }
  27. };
  28. /**
  29. * @param {string} domainSetPath
  30. */
  31. const processLocalDomainSet = async (domainSetPath) => {
  32. for await (const line of readFileByLine(domainSetPath)) {
  33. const parsed = tldts.parse(line, { allowPrivateDomains: true, detectIp: false });
  34. if (parsed.isIp) continue;
  35. if (parsed.isIcann || parsed.isPrivate) {
  36. if (parsed.domain) {
  37. set.add(parsed.domain);
  38. }
  39. continue;
  40. }
  41. if (processLine(line)) {
  42. console.warn('[drop line from domainset]', line);
  43. }
  44. }
  45. };
  46. /**
  47. * @param {string} ruleSetPath
  48. */
  49. const processLocalRuleSet = async (ruleSetPath) => {
  50. for await (const line of readFileByLine(ruleSetPath)) {
  51. if (line.startsWith('DOMAIN-SUFFIX,')) {
  52. addApexDomain(line.replace('DOMAIN-SUFFIX,', ''));
  53. } else if (line.startsWith('DOMAIN,')) {
  54. addApexDomain(line.replace('DOMAIN,', ''));
  55. } else if (line.startsWith('DOMAIN-KEYWORD')) {
  56. keywords.add(escapeRegExp(line.replace('DOMAIN-KEYWORD,', '')));
  57. } else if (line.startsWith('USER-AGENT,') || line.startsWith('PROCESS-NAME,')) {
  58. // do nothing
  59. } else if (processLine(line)) {
  60. console.warn('[drop line from ruleset]', line);
  61. }
  62. }
  63. };
  64. await Promise.all([
  65. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/cdn.conf')),
  66. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/global.conf')),
  67. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/global_plus.conf')),
  68. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/my_proxy.conf')),
  69. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/stream.conf')),
  70. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/telegram.conf')),
  71. processLocalDomainSet(path.resolve(__dirname, '../List/domainset/cdn.conf')),
  72. processLocalDomainSet(path.resolve(__dirname, '../List/domainset/download.conf')),
  73. fsp.mkdir(path.resolve(__dirname, '../List/internal'), { recursive: true })
  74. ]);
  75. return compareAndWriteFile(
  76. [
  77. ...Array.from(set).sort(domainSorter).map(i => `SUFFIX,${i}`),
  78. ...Array.from(keywords).sort().map(i => `REGEX,${i}`)
  79. ],
  80. path.resolve(__dirname, '../List/internal/cdn.txt')
  81. );
  82. });
  83. module.exports.buildInternalCDNDomains = buildInternalCDNDomains;
  84. if (require.main === module) {
  85. buildInternalCDNDomains();
  86. }