tools-lum-apex-domains.ts 2.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748
  1. import { fetchRemoteTextByLine } from './lib/fetch-text-by-line';
  2. import tldts from 'tldts-experimental';
  3. import { HostnameSmolTrie } from 'hntrie/smol';
  4. import { domainToASCII } from 'node:url';
  5. import path from 'node:path';
  6. import { SOURCE_DIR } from './constants/dir';
  7. import runAgainstSourceFile from './lib/run-against-source-file';
  8. (async () => {
  9. const lines1 = await Array.fromAsync(await fetchRemoteTextByLine('https://raw.githubusercontent.com/durablenapkin/block/master/luminati.txt', true));
  10. const lines2 = await Array.fromAsync(await fetchRemoteTextByLine('https://raw.githubusercontent.com/durablenapkin/block/master/tvstream.txt', true));
  11. const trie = new HostnameSmolTrie();
  12. lines1.forEach((line) => {
  13. const apexDomain = tldts.getDomain(line.slice(8));
  14. if (apexDomain) {
  15. trie.add(apexDomain);
  16. }
  17. });
  18. lines2.forEach((line) => {
  19. const apexDomain = tldts.getDomain(line.slice(8));
  20. if (apexDomain) {
  21. trie.add(apexDomain);
  22. }
  23. });
  24. const dataFromDuckDuckGo = await fetch('https://raw.githubusercontent.com/duckduckgo/tracker-radar/92e086ce38a8a88c964ed0184e5277ec1d5c8038/entities/Bright%20Data%20Ltd..json').then((res) => res.json());
  25. if (typeof dataFromDuckDuckGo === 'object' && dataFromDuckDuckGo !== null && 'properties' in dataFromDuckDuckGo && Array.isArray(dataFromDuckDuckGo.properties)) {
  26. dataFromDuckDuckGo.properties.forEach((prop) => {
  27. trie.add(prop);
  28. });
  29. }
  30. await runAgainstSourceFile(path.join(SOURCE_DIR, 'domainset', 'reject.conf'), (domain, includeAllSubDomain) => {
  31. trie.whitelist(includeAllSubDomain ? '.' + domain : domain);
  32. }, 'domainset');
  33. await runAgainstSourceFile(path.join(SOURCE_DIR, 'non_ip', 'reject.conf'), (domain, includeAllSubDomain) => {
  34. trie.whitelist(includeAllSubDomain ? '.' + domain : domain);
  35. }, 'ruleset');
  36. const dump: string[] = [];
  37. trie.dump((rawDomain) => {
  38. const domain = domainToASCII(rawDomain);
  39. if (domain) dump.push(domain);
  40. });
  41. console.log(dump.map(i => '.' + i).join('\n'));
  42. })();