tools-dedupe-src.ts 3.4 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091
  1. import { fdir as Fdir } from 'fdir';
  2. import path from 'node:path';
  3. import fsp from 'node:fs/promises';
  4. import { SOURCE_DIR } from './constants/dir';
  5. import { readFileByLine } from './lib/fetch-text-by-line';
  6. import { processLine } from './lib/process-line';
  7. import { HostnameSmolTrie, HostnameTrie } from './lib/trie';
  8. import { task } from './trace';
  9. const ENFORCED_WHITELIST = [
  10. 'hola.sk',
  11. 'hola.org',
  12. 'hola-shopping.com',
  13. 'mynextphone.io',
  14. 'iadmatapk.nosdn.127.net',
  15. 'httpdns.bilivideo.com',
  16. 'httpdns-v6.gslb.yy.com',
  17. 'twemoji.maxcdn.com',
  18. 'samsungcloudsolution.com',
  19. 'samsungcloudsolution.net',
  20. 'samsungqbe.com'
  21. ];
  22. const WHITELIST: string[] = ['.dxdhd.com', '.tokto-motion.net', '.hola-shopping.com', '.luxxeeu.com', '.newzgames.com', '.hola.com.sg', 'pengtu.cc', '.cdn-js-query.com', 'samsungcloudsolution.net', 'samsungcloudsolution.com', 'static.estebull.com', '.drawservant.com', '.enjoy7plains.xyz', '.zmfindyourhalf.top', '.mineblocks.eu', '.cointaft.com', '.chain-pool.com', '.lamby-crypto.com', '.grftpool.com', '.onebtcplace.com', '.pepecore.com', '.punchsub.net', '.imzlabs.net', '.datapaw.net', '.smpool.net', '.yetimining.net', '.igrid.org', '.50centfreedom.us', '.cyg2016.xyz', '.easypool.xyz', '.arhash.xyz', '.enviromint.xyz', '.pool.space', '.anomp.cc', '.bitconnectpool.co', '.cryptopool.space', '.automatix.to', '.coolmine.to', '.coolpool.to', '.dpool.to', '.template-download.to', '.aurum7.to', '.sunpool.to', '.speedpool.to', '.cfcnet.to', '.pool.do', '.pool.bit34.com', '.eos.zhizhu.to', '.mubicdn.com', 'cdn.fastmediaing.com', '.webinfcdn.com', '.aosikaimage.com'];
  23. task(require.main === module, __filename)(async (span) => {
  24. const files = await span.traceChildAsync('crawl thru all files', () => new Fdir()
  25. .withFullPaths()
  26. .filter((filepath, isDirectory) => {
  27. if (isDirectory) return true;
  28. const extname = path.extname(filepath);
  29. return extname !== '.js' && extname !== '.ts';
  30. })
  31. .crawl(SOURCE_DIR)
  32. .withPromise());
  33. const whiteTrie = span.traceChildSync('build whitelist trie', () => {
  34. const trie = new HostnameSmolTrie(WHITELIST);
  35. ENFORCED_WHITELIST.forEach((item) => trie.whitelist(item));
  36. return trie;
  37. });
  38. await Promise.all(files.map(file => span.traceChildAsync('dedupe ' + file, () => dedupeFile(file, whiteTrie))));
  39. });
  40. async function dedupeFile(file: string, whitelist: HostnameSmolTrie) {
  41. const result: string[] = [];
  42. const trie = new HostnameTrie();
  43. for await (const l of readFileByLine(file)) {
  44. const line = processLine(l);
  45. if (!line) {
  46. if (l.startsWith('# $ skip_dedupe_src')) {
  47. return;
  48. }
  49. result.push(l); // keep all comments and blank lines
  50. continue;
  51. }
  52. if (trie.has(line)) {
  53. continue; // drop duplicate
  54. }
  55. if (whitelist.has(line)) {
  56. continue; // drop whitelisted items
  57. }
  58. trie.add(line);
  59. result.push(line);
  60. }
  61. return fsp.writeFile(file, result.join('\n') + '\n');
  62. }
  63. // function isDomainSuffix(whiteItem: string, incomingItem: string) {
  64. // const whiteIncludeDomain = whiteItem[0] === '.';
  65. // whiteItem = whiteItem[0] === '.' ? whiteItem.slice(1) : whiteItem;
  66. // if (whiteItem === incomingItem) {
  67. // return true; // as long as exact match, we don't care if subdomain is included or not
  68. // }
  69. // if (whiteIncludeDomain) {
  70. // return incomingItem.endsWith('.' + whiteItem);
  71. // }
  72. // return false;
  73. // }