build-internal-cdn-rules.js 3.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100
  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. // console.log({ line });
  34. const parsed = tldts.parse(line, { allowPrivateDomains: true, detectIp: false });
  35. if (parsed.isIp) continue;
  36. if (parsed.isIcann || parsed.isPrivate) {
  37. if (parsed.domain) {
  38. set.add(parsed.domain);
  39. }
  40. continue;
  41. }
  42. if (processLine(line)) {
  43. console.warn('[drop line from domainset]', line);
  44. }
  45. }
  46. };
  47. /**
  48. * @param {string} ruleSetPath
  49. */
  50. const processLocalRuleSet = async (ruleSetPath) => {
  51. for await (const line of readFileByLine(ruleSetPath)) {
  52. if (line.startsWith('DOMAIN-SUFFIX,')) {
  53. addApexDomain(line.replace('DOMAIN-SUFFIX,', ''));
  54. } else if (line.startsWith('DOMAIN,')) {
  55. addApexDomain(line.replace('DOMAIN,', ''));
  56. } else if (line.startsWith('DOMAIN-KEYWORD')) {
  57. keywords.add(escapeRegExp(line.replace('DOMAIN-KEYWORD,', '')));
  58. } else if (line.startsWith('USER-AGENT,') || line.startsWith('PROCESS-NAME,')) {
  59. // do nothing
  60. } else if (processLine(line)) {
  61. console.warn('[drop line from ruleset]', line);
  62. }
  63. }
  64. };
  65. await Promise.all([
  66. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/cdn.conf')),
  67. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/global.conf')),
  68. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/global_plus.conf')),
  69. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/my_proxy.conf')),
  70. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/stream.conf')),
  71. processLocalRuleSet(path.resolve(__dirname, '../List/non_ip/telegram.conf')),
  72. processLocalDomainSet(path.resolve(__dirname, '../List/domainset/cdn.conf')),
  73. processLocalDomainSet(path.resolve(__dirname, '../List/domainset/download.conf')),
  74. fsp.mkdir(path.resolve(__dirname, '../List/internal'), { recursive: true })
  75. ]);
  76. return compareAndWriteFile(
  77. [
  78. ...Array.from(set).sort(domainSorter).map(i => `SUFFIX,${i}`),
  79. ...Array.from(keywords).sort().map(i => `REGEX,${i}`)
  80. ],
  81. path.resolve(__dirname, '../List/internal/cdn.txt')
  82. );
  83. });
  84. module.exports.buildInternalCDNDomains = buildInternalCDNDomains;
  85. if (import.meta.main) {
  86. buildInternalCDNDomains();
  87. }