tools-dedupe-src.ts 4.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103
  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 } 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. 'ntp.api.bz',
  22. 'cdn.tuk.dev',
  23. 'vocadb-analytics.fly.dev',
  24. 'img.vim-cn.com'
  25. ];
  26. const DEDUPE_LIST: string[] = ['127.atlas.skk.moe', 'ntp.api.bz', 'httpdns.bilivideo.com', 'dns.qiyipic.iqiyi.com', 'cdn.graph.office.net', 'dns.iqiyi.com', 'img.vim-cn.com', 'image.westinyou.com', 'edge1.certona.net', 'certona.gap.com', 'yep.video.yahoo.com', 'static.opensea.io', 'shopify.cleverecommerce.com', 'tile.mapzen.com', 'cdn.cracked.sh', 'images.idgesg.net', 'drive.massgrave.dev', 'alt.idgesg.net', 'mirror.ghproxy.com', 'mirror.nl.datapacket.com', 'mirror.anigil.com', 'mirror.nus.edu.sg', 'mirror.timkevin.us', 'mirrors.nic.cz', 'cpan.tetaneutral.net', 'mirror.datapacket.com', 'client.hikarifield.co.jp', 'a.macked.app', 'apache.tt.co.kr', 'fm.p0y.cn', 'iyes.youku.com', 'ad.api.mobile.youku.com', 'c.yes.youku.com', 'ad.jamster.co.uk', 'fumiad.dxys.pro', 'ad.leadboltapps.net', 'ems.cp12.wasu.tv', 'creative1cdn.mobfox.com', 'mycommerce.akamaized.net', 'js-cdn.blockchair.io', 'loutre.blockchair.io', 'static.namebeta.com', 'fs2.onlyhentaistuff.com', 'file.izanmei.net', 'play.xiaoh.ai', 'file.xiaohai.ai', 'tiles.wmflabs.org', 'image.stheadline.com', 'vod.jfly.xyz', 'assets.wikiwand.com', 'cdn.wikiwand.com', 'cloudfront.codeproject.com', 'assets-cdn.anh.moe', 'media.d.tube', 'media.remax-prod.eng.remax.tech', 'static-landing.probiplacehold.cot.com'];
  27. task(require.main === module, __filename)(async (span) => {
  28. const files = await span.traceChildAsync('crawl thru all files', () => new Fdir()
  29. .withFullPaths()
  30. .filter((filepath, isDirectory) => {
  31. if (isDirectory) return true;
  32. const extname = path.extname(filepath);
  33. return extname !== '.js' && extname !== '.ts';
  34. })
  35. .crawl(SOURCE_DIR)
  36. .withPromise());
  37. const whiteTrie = span.traceChildSync('build whitelist trie', () => {
  38. const trie = new HostnameSmolTrie(DEDUPE_LIST);
  39. ENFORCED_WHITELIST.forEach((item) => trie.whitelist(item));
  40. return trie;
  41. });
  42. await Promise.all(files.map(file => span.traceChildAsync('dedupe ' + file, () => dedupeFile(file, whiteTrie))));
  43. });
  44. async function dedupeFile(file: string, whitelist: HostnameSmolTrie) {
  45. const result: string[] = [];
  46. const trie = new HostnameSmolTrie();
  47. let line: string | null = '';
  48. // eslint-disable-next-line @typescript-eslint/unbound-method -- .call
  49. let trieHasOrContains = HostnameSmolTrie.prototype.has;
  50. for await (const l of readFileByLine(file)) {
  51. line = processLine(l);
  52. if (!line) {
  53. if (l.startsWith('# $ skip_dedupe_src')) {
  54. return;
  55. }
  56. if (l.startsWith('# $ dedupe_use_trie_contains')) {
  57. // eslint-disable-next-line @typescript-eslint/unbound-method -- .call
  58. trieHasOrContains = HostnameSmolTrie.prototype.contains;
  59. }
  60. result.push(l); // keep all comments and blank lines
  61. continue;
  62. }
  63. if (trieHasOrContains.call(trie, line)) {
  64. continue; // drop duplicate
  65. }
  66. if (whitelist.has(line)) {
  67. continue; // drop whitelisted items
  68. }
  69. trie.add(line);
  70. result.push(line);
  71. }
  72. return fsp.writeFile(file, result.join('\n') + '\n');
  73. }
  74. // function isDomainSuffix(whiteItem: string, incomingItem: string) {
  75. // const whiteIncludeDomain = whiteItem[0] === '.';
  76. // whiteItem = whiteItem[0] === '.' ? whiteItem.slice(1) : whiteItem;
  77. // if (whiteItem === incomingItem) {
  78. // return true; // as long as exact match, we don't care if subdomain is included or not
  79. // }
  80. // if (whiteIncludeDomain) {
  81. // return incomingItem.endsWith('.' + whiteItem);
  82. // }
  83. // return false;
  84. // }