build-reject-domainset.js 8.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241
  1. const { promises: fsPromises } = require('fs');
  2. const { resolve: pathResolve } = require('path');
  3. const Piscina = require('piscina');
  4. const { processHosts, processFilterRules } = require('./lib/parse-filter');
  5. const cpuCount = require('os').cpus().length;
  6. const { isCI } = require('ci-info');
  7. const threads = isCI ? cpuCount : cpuCount / 2;
  8. (async () => {
  9. /** @type Set<string> */
  10. const domainSets = new Set();
  11. // Parse from remote hosts & domain lists
  12. (await Promise.all([
  13. processHosts('https://pgl.yoyo.org/adservers/serverlist.php?hostformat=hosts&showintro=0&mimetype=plaintext', true),
  14. processHosts('https://raw.githubusercontent.com/hoshsadiq/adblock-nocoin-list/master/hosts.txt'),
  15. processHosts('https://raw.githubusercontent.com/crazy-max/WindowsSpyBlocker/master/data/hosts/spy.txt')
  16. ])).forEach(hosts => {
  17. hosts.forEach(host => {
  18. if (host) {
  19. domainSets.add(host);
  20. }
  21. });
  22. });
  23. let previousSize = domainSets.size;
  24. console.log(`Import ${previousSize} rules from hosts files!`);
  25. await fsPromises.readFile(pathResolve(__dirname, '../List/domainset/reject_sukka.conf'), { encoding: 'utf-8' }).then(data => {
  26. data.split('\n').forEach(line => {
  27. const trimmed = line.trim();
  28. if (
  29. line.startsWith('#')
  30. || line.startsWith(' ')
  31. || line.startsWith('\r')
  32. || line.startsWith('\n')
  33. || trimmed === ''
  34. ) {
  35. return;
  36. }
  37. /* if (domainSets.has(line) || domainSets.has(`.${line}`)) {
  38. console.warn(`|${line}| is already in the list!`);
  39. } */
  40. domainSets.add(trimmed);
  41. });
  42. });
  43. previousSize = domainSets.size - previousSize;
  44. console.log(`Import ${previousSize} rules from reject_sukka.conf!`);
  45. // Parse from AdGuard Filters
  46. /** @type Set<string> */
  47. const filterRuleWhitelistDomainSets = new Set([
  48. 'localhost',
  49. 'broadcasthost',
  50. 'ip6-loopback',
  51. 'ip6-localnet',
  52. 'ip6-mcastprefix',
  53. 'ip6-allnodes',
  54. 'ip6-allrouters',
  55. 'ip6-allhosts',
  56. 'mcastprefix',
  57. 'analytics.google.com',
  58. 'msa.cdn.mediaset.net', // Added manually using DOMAIN-KEYWORDS
  59. 'cloud.answerhub.com',
  60. 'ae01.alicdn.com',
  61. 'whoami.akamai.net',
  62. 'whoami.ds.akahelp.net',
  63. 'pxlk9.net.', // This one is malformed from EasyList, which I will manually add instead
  64. 'instant.page', // No, it doesn't violate anyone's privacy. I will whitelist it
  65. 'piwik.pro',
  66. 'mixpanel.com',
  67. 'cdn.mxpnl.com',
  68. 'heapanalytics.com',
  69. 'segment.com',
  70. 'segmentify.com',
  71. 't.co', // pgl yoyo add t.co to the blacklist
  72. 'survicate.com', // AdGuardDNSFilter
  73. 'perfops.io', // AdGuardDNSFilter
  74. 'd2axgrpnciinw7.cloudfront.net' // ADGuardDNSFilter
  75. ]);
  76. (await Promise.all([
  77. // Easy List
  78. 'https://easylist.to/easylist/easylist.txt',
  79. // AdGuard DNS Filter
  80. 'https://adguardteam.github.io/AdGuardSDNSFilter/Filters/filter.txt',
  81. // uBlock Origin Filter List
  82. 'https://ublockorigin.github.io/uAssetsCDN/filters/filters.txt',
  83. 'https://ublockorigin.github.io/uAssetsCDN/filters/filters-2020.txt',
  84. 'https://ublockorigin.github.io/uAssetsCDN/filters/filters-2021.txt',
  85. 'https://ublockorigin.github.io/uAssetsCDN/filters/filters-2022.txt',
  86. // uBlock Origin Badware Risk List
  87. 'https://ublockorigin.github.io/uAssets/filters/badware.txt',
  88. // uBlock Origin Privacy List
  89. 'https://ublockorigin.github.io/uAssets/filters/privacy.txt',
  90. // uBlock Origin Resource Abuse
  91. 'https://ublockorigin.github.io/uAssets/filters/resource-abuse.txt',
  92. // uBlock Origin Unbreak
  93. 'https://ublockorigin.github.io/uAssets/filters/unbreak.txt',
  94. // AdGuard Base Filter
  95. 'https://filters.adtidy.org/extension/ublock/filters/2_without_easylist.txt',
  96. // AdGuard Mobile AD
  97. 'https://filters.adtidy.org/extension/ublock/filters/11.txt',
  98. // AdGuard Tracking Protection
  99. 'https://filters.adtidy.org/extension/ublock/filters/3.txt',
  100. // AdGuard Japanese filter
  101. 'https://filters.adtidy.org/extension/ublock/filters/7.txt',
  102. // AdGuard Chinese filter (EasyList China + AdGuard Chinese filter)
  103. 'https://filters.adtidy.org/extension/ublock/filters/224.txt',
  104. // Easy Privacy
  105. 'https://easylist.to/easylist/easyprivacy.txt',
  106. // Curben's Malware Online UrlHaus
  107. 'https://malware-filter.gitlab.io/malware-filter/urlhaus-filter-agh-online.txt',
  108. // Curben's Phishing Online Filter
  109. 'https://malware-filter.gitlab.io/malware-filter/phishing-filter-agh.txt',
  110. // Curben's PUP List
  111. 'https://malware-filter.gitlab.io/malware-filter/pup-filter-agh.txt',
  112. // GameConsoleAdblockList
  113. 'https://raw.githubusercontent.com/DandelionSprout/adfilt/master/GameConsoleAdblockList.txt',
  114. // PiHoleBlocklist
  115. 'https://raw.githubusercontent.com/Perflyst/PiHoleBlocklist/master/SmartTV-AGH.txt',
  116. ].map(processFilterRules))).forEach(({ white, black }) => {
  117. white.forEach(i => filterRuleWhitelistDomainSets.add(i));
  118. black.forEach(i => domainSets.add(i));
  119. });
  120. previousSize = domainSets.size - previousSize;
  121. console.log(`Import ${previousSize} rules from adguard filters!`);
  122. // Read DOMAIN Keyword
  123. const domainKeywordsSet = new Set();
  124. const domainSuffixSet = new Set();
  125. await fsPromises.readFile(pathResolve(__dirname, '../List/non_ip/reject.conf'), { encoding: 'utf-8' }).then(data => {
  126. data.split('\n').forEach(line => {
  127. if (line.startsWith('DOMAIN-KEYWORD')) {
  128. const [, ...keywords] = line.split(',');
  129. domainKeywordsSet.add(keywords.join(',').trim());
  130. } else if (line.startsWith('DOMAIN-SUFFIX')) {
  131. const [, ...keywords] = line.split(',');
  132. domainSuffixSet.add(keywords.join(',').trim());
  133. }
  134. });
  135. });
  136. console.log(`Import ${domainKeywordsSet.size} black keywords and ${domainSuffixSet.size} black suffixes!`);
  137. previousSize = domainSets.size;
  138. // Dedupe domainSets
  139. console.log(`Start deduping! (${previousSize})`);
  140. const toBeRemoved = new Set();
  141. for (const domain of domainSets) {
  142. let isTobeRemoved = false;
  143. for (const keyword of domainKeywordsSet) {
  144. if (domain.includes(keyword) || keyword.includes(domain)) {
  145. isTobeRemoved = true;
  146. break;
  147. }
  148. }
  149. if (!isTobeRemoved) {
  150. for (const suffix of domainSuffixSet) {
  151. if (domain.endsWith(suffix)) {
  152. isTobeRemoved = true;
  153. break;
  154. }
  155. }
  156. }
  157. if (!isTobeRemoved) {
  158. for (const white of filterRuleWhitelistDomainSets) {
  159. if (domain.includes(white) || white.includes(domain)) {
  160. isTobeRemoved = true;
  161. break;
  162. }
  163. }
  164. }
  165. if (isTobeRemoved) {
  166. toBeRemoved.add(domain);
  167. }
  168. }
  169. toBeRemoved.forEach((removed) => {
  170. domainSets.delete(removed)
  171. });
  172. // Dedupe domainSets
  173. console.log(`Deduped ${previousSize - domainSets.size} from black keywords and suffixes!`);
  174. previousSize = domainSets.size;
  175. // Dedupe domainSets
  176. console.log(`Start deduping! (${previousSize})`);
  177. const piscina = new Piscina({
  178. filename: pathResolve(__dirname, 'worker/build-reject-domainset-worker.js'),
  179. workerData: [...domainSets]
  180. });
  181. console.log(`Launching ${threads} threads...`)
  182. const tasksArray = Array.from(domainSets)
  183. .reduce((result, element, index) => {
  184. const chunk = index % threads;
  185. result[chunk] ??= [];
  186. result[chunk].push(element);
  187. return result;
  188. }, []);
  189. (await Promise.all(
  190. Array.from(domainSets)
  191. .reduce((result, element, index) => {
  192. const chunk = index % threads;
  193. result[chunk] ??= [];
  194. result[chunk].push(element);
  195. return result;
  196. }, [])
  197. .map(chunk => piscina.run(
  198. { chunk }
  199. ))
  200. )).forEach((result, taskIndex) => {
  201. const chunk = tasksArray[taskIndex];
  202. result.forEach((value, index) => {
  203. if (value === 1) {
  204. domainSets.delete(chunk[index])
  205. }
  206. })
  207. });
  208. console.log(`Deduped ${previousSize - domainSets.size} rules!`);
  209. return fsPromises.writeFile(
  210. pathResolve(__dirname, '../List/domainset/reject.conf'),
  211. `${[...domainSets].join('\n')}\n`,
  212. { encoding: 'utf-8' });
  213. })();