build-reject-domainset.ts 9.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187
  1. // @ts-check
  2. import path from 'node:path';
  3. import process from 'node:process';
  4. import { processHostsWithPreload } from './lib/parse-filter/hosts';
  5. import { processDomainListsWithPreload } from './lib/parse-filter/domainlists';
  6. import { processFilterRulesWithPreload } from './lib/parse-filter/filters';
  7. import { HOSTS, ADGUARD_FILTERS, PREDEFINED_WHITELIST, DOMAIN_LISTS, HOSTS_EXTRA, DOMAIN_LISTS_EXTRA, ADGUARD_FILTERS_EXTRA, ADGUARD_FILTERS_WHITELIST, PHISHING_HOSTS_EXTRA, PHISHING_DOMAIN_LISTS_EXTRA } from './constants/reject-data-source';
  8. import { readFileIntoProcessedArray } from './lib/fetch-text-by-line';
  9. import { task } from './trace';
  10. // tldts-experimental is way faster than tldts, but very little bit inaccurate
  11. // (since it is hashes based). But the result is still deterministic, which is
  12. // enough when creating a simple stat of reject hosts.
  13. import { SHARED_DESCRIPTION } from './constants/description';
  14. import { addArrayElementsToSet } from 'foxts/add-array-elements-to-set';
  15. import { OUTPUT_INTERNAL_DIR, SOURCE_DIR } from './constants/dir';
  16. import { DomainsetOutput } from './lib/rules/domainset';
  17. import { foundDebugDomain } from './lib/parse-filter/shared';
  18. import { AdGuardHomeOutput } from './lib/rules/domainset';
  19. import { getPhishingDomains } from './lib/get-phishing-domains';
  20. const readLocalRejectDomainsetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'domainset/reject.conf'));
  21. const readLocalRejectExtraDomainsetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'domainset/reject_extra.conf'));
  22. const readLocalRejectRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/reject.conf'));
  23. const readLocalRejectDropRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/reject-drop.conf'));
  24. const readLocalRejectNoDropRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/reject-no-drop.conf'));
  25. const readLocalMyRejectRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/my_reject.conf'));
  26. const hostsDownloads = HOSTS.map(entry => processHostsWithPreload(...entry));
  27. const hostsExtraDownloads = HOSTS_EXTRA.map(entry => processHostsWithPreload(...entry));
  28. const domainListsDownloads = DOMAIN_LISTS.map(entry => processDomainListsWithPreload(...entry));
  29. const domainListsExtraDownloads = DOMAIN_LISTS_EXTRA.map(entry => processDomainListsWithPreload(...entry));
  30. const adguardFiltersDownloads = ADGUARD_FILTERS.map(entry => processFilterRulesWithPreload(...entry));
  31. const adguardFiltersExtraDownloads = ADGUARD_FILTERS_EXTRA.map(entry => processFilterRulesWithPreload(...entry));
  32. const adguardFiltersWhitelistsDownloads = ADGUARD_FILTERS_WHITELIST.map(entry => processFilterRulesWithPreload(...entry));
  33. export const buildRejectDomainSet = task(require.main === module, __filename)(async (span) => {
  34. const rejectBaseDescription = [
  35. ...SHARED_DESCRIPTION,
  36. '',
  37. 'The domainset supports AD blocking, tracking protection, privacy protection, anti-mining',
  38. '',
  39. 'Build from:',
  40. ...HOSTS.map(host => ` - ${host[0]}`),
  41. ...DOMAIN_LISTS.map(domainList => ` - ${domainList[0]}`),
  42. ...ADGUARD_FILTERS.map(filter => ` - ${Array.isArray(filter) ? filter[0] : filter}`)
  43. ];
  44. const rejectOutput = new DomainsetOutput(span, 'reject')
  45. .withTitle('Sukka\'s Ruleset - Reject Base')
  46. .withDescription(rejectBaseDescription);
  47. const rejectExtraOutput = new DomainsetOutput(span, 'reject_extra')
  48. .withTitle('Sukka\'s Ruleset - Reject Extra')
  49. .withDescription([
  50. ...SHARED_DESCRIPTION,
  51. '',
  52. 'The domainset supports AD blocking, tracking protection, privacy protection, anti-mining',
  53. '',
  54. 'Build from:',
  55. ...HOSTS_EXTRA.map(host => ` - ${host[0]}`),
  56. ...DOMAIN_LISTS_EXTRA.map(domainList => ` - ${domainList[0]}`),
  57. ...ADGUARD_FILTERS_EXTRA.map(filter => ` - ${filter[0]}`)
  58. ]);
  59. const rejectPhisingOutput = new DomainsetOutput(span, 'reject_phishing')
  60. .withTitle('Sukka\'s Ruleset - Reject Phishing')
  61. .withDescription([
  62. ...SHARED_DESCRIPTION,
  63. '',
  64. 'The domainset is specifically designed for anti-phishing',
  65. '',
  66. 'Build from:',
  67. ...PHISHING_HOSTS_EXTRA.map(host => ` - ${host[0]}`),
  68. ...PHISHING_DOMAIN_LISTS_EXTRA.map(domainList => ` - ${domainList[0]}`)
  69. ]);
  70. const appendArrayToRejectOutput = rejectOutput.addFromDomainset.bind(rejectOutput);
  71. const appendArrayToRejectExtraOutput = rejectExtraOutput.addFromDomainset.bind(rejectExtraOutput);
  72. /** Whitelists */
  73. const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
  74. // Parse from AdGuard Filters
  75. await span
  76. .traceChild('download and process hosts / adblock filter rules')
  77. .traceAsyncFn((childSpan) => Promise.all([
  78. // Dedupe domainSets
  79. // Collect DOMAIN, DOMAIN-SUFFIX, and DOMAIN-KEYWORD from non_ip/reject.conf for deduplication
  80. // DOMAIN-WILDCARD is not really useful for deduplication, it is only included in AdGuardHome output
  81. // It is faster to add base than add others first then whitelist
  82. rejectOutput.addFromRuleset(readLocalRejectRulesetPromise),
  83. rejectExtraOutput.addFromRuleset(readLocalRejectRulesetPromise),
  84. rejectOutput.addFromDomainset(readLocalRejectDomainsetPromise),
  85. rejectExtraOutput.addFromDomainset(readLocalRejectDomainsetPromise),
  86. rejectPhisingOutput.addFromDomainset(readLocalRejectDomainsetPromise),
  87. rejectExtraOutput.addFromDomainset(readLocalRejectExtraDomainsetPromise),
  88. // Parse from remote hosts & domain lists
  89. hostsDownloads.map(task => task(childSpan).then(appendArrayToRejectOutput)),
  90. hostsExtraDownloads.map(task => task(childSpan).then(appendArrayToRejectExtraOutput)),
  91. domainListsDownloads.map(task => task(childSpan).then(appendArrayToRejectOutput)),
  92. domainListsExtraDownloads.map(task => task(childSpan).then(appendArrayToRejectExtraOutput)),
  93. rejectPhisingOutput.addFromDomainset(getPhishingDomains(childSpan)),
  94. adguardFiltersDownloads.map(
  95. task => task(childSpan).then(({ whiteDomains, whiteDomainSuffixes, blackDomains, blackDomainSuffixes }) => {
  96. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomains);
  97. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomainSuffixes, suffix => '.' + suffix);
  98. rejectOutput.bulkAddDomain(blackDomains);
  99. rejectOutput.bulkAddDomainSuffix(blackDomainSuffixes);
  100. })
  101. ),
  102. adguardFiltersExtraDownloads.map(
  103. task => task(childSpan).then(({ whiteDomains, whiteDomainSuffixes, blackDomains, blackDomainSuffixes }) => {
  104. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomains);
  105. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomainSuffixes, suffix => '.' + suffix);
  106. rejectExtraOutput.bulkAddDomain(blackDomains);
  107. rejectExtraOutput.bulkAddDomainSuffix(blackDomainSuffixes);
  108. })
  109. ),
  110. adguardFiltersWhitelistsDownloads.map(
  111. task => task(childSpan).then(({ whiteDomains, whiteDomainSuffixes, blackDomains, blackDomainSuffixes }) => {
  112. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomains);
  113. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomainSuffixes, suffix => '.' + suffix);
  114. addArrayElementsToSet(filterRuleWhitelistDomainSets, blackDomains);
  115. addArrayElementsToSet(filterRuleWhitelistDomainSets, blackDomainSuffixes, suffix => '.' + suffix);
  116. })
  117. )
  118. ].flat()));
  119. if (foundDebugDomain.value) {
  120. // eslint-disable-next-line sukka/unicorn/no-process-exit -- cli App
  121. process.exit(1);
  122. }
  123. await Promise.all([
  124. rejectOutput.done(),
  125. rejectExtraOutput.done(),
  126. rejectPhisingOutput.done()
  127. ]);
  128. // whitelist
  129. span.traceChildSync('whitelist', () => {
  130. for (const domain of filterRuleWhitelistDomainSets) {
  131. rejectOutput.whitelistDomain(domain);
  132. rejectExtraOutput.whitelistDomain(domain);
  133. rejectPhisingOutput.whitelistDomain(domain);
  134. }
  135. rejectOutput.domainTrie.dump(rejectExtraOutput.whitelistDomain.bind(rejectExtraOutput));
  136. rejectOutput.domainTrie.dump(rejectPhisingOutput.whitelistDomain.bind(rejectPhisingOutput));
  137. });
  138. await Promise.all([
  139. rejectOutput.write(),
  140. rejectExtraOutput.write(),
  141. rejectPhisingOutput.write()
  142. ]);
  143. // we are going to re-use rejectOutput's domainTrie and mutate it
  144. // so we must wait until we write rejectOutput to disk after we can mutate its trie
  145. const rejectOutputAdGuardHome = new AdGuardHomeOutput(span, 'reject-adguardhome', OUTPUT_INTERNAL_DIR)
  146. .withTitle('Sukka\'s Ruleset - Blocklist for AdGuardHome')
  147. .withDescription([
  148. 'The AdGuardHome ruleset supports AD blocking, tracking protection, privacy protection, anti-mining'
  149. ]);
  150. rejectOutputAdGuardHome.domainTrie = rejectOutput.domainTrie;
  151. await rejectOutputAdGuardHome
  152. .addFromRuleset(readLocalMyRejectRulesetPromise)
  153. .addFromRuleset(readLocalRejectRulesetPromise)
  154. .addFromRuleset(readLocalRejectDropRulesetPromise)
  155. .addFromRuleset(readLocalRejectNoDropRulesetPromise)
  156. .addFromDomainset(readLocalRejectExtraDomainsetPromise)
  157. // .
  158. .write();
  159. });