build-reject-domainset.ts 8.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169
  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, PHISHING_DOMAIN_LISTS_EXTRA, ADGUARD_FILTERS_WHITELIST } 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 { getPhishingDomains } from './lib/get-phishing-domains';
  15. import { addArrayElementsToSet } from 'foxts/add-array-elements-to-set';
  16. import { OUTPUT_INTERNAL_DIR, SOURCE_DIR } from './constants/dir';
  17. import { DomainsetOutput } from './lib/create-file';
  18. import { foundDebugDomain } from './lib/parse-filter/shared';
  19. import { AdGuardHome } from './lib/writing-strategy/adguardhome';
  20. import { FileOutput } from './lib/rules/base';
  21. const readLocalRejectDomainsetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'domainset/reject_sukka.conf'));
  22. const readLocalRejectExtraDomainsetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'domainset/reject_sukka_extra.conf'));
  23. const readLocalRejectRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/reject.conf'));
  24. const readLocalRejectDropRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/reject-drop.conf'));
  25. const readLocalRejectNoDropRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/reject-no-drop.conf'));
  26. const readLocalMyRejectRulesetPromise = readFileIntoProcessedArray(path.join(SOURCE_DIR, 'non_ip/my_reject.conf'));
  27. const hostsDownloads = HOSTS.map(entry => processHostsWithPreload(...entry));
  28. const hostsExtraDownloads = HOSTS_EXTRA.map(entry => processHostsWithPreload(...entry));
  29. const domainListsDownloads = DOMAIN_LISTS.map(entry => processDomainListsWithPreload(...entry));
  30. const domainListsExtraDownloads = DOMAIN_LISTS_EXTRA.map(entry => processDomainListsWithPreload(...entry));
  31. const adguardFiltersDownloads = ADGUARD_FILTERS.map(entry => processFilterRulesWithPreload(...entry));
  32. const adguardFiltersExtraDownloads = ADGUARD_FILTERS_EXTRA.map(entry => processFilterRulesWithPreload(...entry));
  33. const adguardFiltersWhitelistsDownloads = ADGUARD_FILTERS_WHITELIST.map(entry => processFilterRulesWithPreload(...entry));
  34. export const buildRejectDomainSet = task(require.main === module, __filename)(async (span) => {
  35. const rejectBaseDescription = [
  36. ...SHARED_DESCRIPTION,
  37. '',
  38. 'The domainset supports AD blocking, tracking protection, privacy protection, anti-phishing, anti-mining',
  39. '',
  40. 'Build from:',
  41. ...HOSTS.map(host => ` - ${host[0]}`),
  42. ...DOMAIN_LISTS.map(domainList => ` - ${domainList[0]}`),
  43. ...ADGUARD_FILTERS.map(filter => ` - ${Array.isArray(filter) ? filter[0] : filter}`)
  44. ];
  45. const rejectOutput = new DomainsetOutput(span, 'reject')
  46. .withTitle('Sukka\'s Ruleset - Reject Base')
  47. .withDescription(rejectBaseDescription);
  48. const rejectExtraOutput = new DomainsetOutput(span, 'reject_extra')
  49. .withTitle('Sukka\'s Ruleset - Reject Extra')
  50. .withDescription([
  51. ...SHARED_DESCRIPTION,
  52. '',
  53. 'The domainset supports AD blocking, tracking protection, privacy protection, anti-phishing, anti-mining',
  54. '',
  55. 'Build from:',
  56. ...HOSTS_EXTRA.map(host => ` - ${host[0]}`),
  57. ...DOMAIN_LISTS_EXTRA.map(domainList => ` - ${domainList[0]}`),
  58. ...ADGUARD_FILTERS_EXTRA.map(filter => ` - ${Array.isArray(filter) ? filter[0] : filter}`),
  59. ...PHISHING_DOMAIN_LISTS_EXTRA.map(domainList => ` - ${domainList[0]}`)
  60. ]);
  61. const appendArrayToRejectOutput = rejectOutput.addFromDomainset.bind(rejectOutput);
  62. const appendArrayToRejectExtraOutput = rejectExtraOutput.addFromDomainset.bind(rejectExtraOutput);
  63. /** Whitelists */
  64. const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
  65. // Parse from AdGuard Filters
  66. await span
  67. .traceChild('download and process hosts / adblock filter rules')
  68. .traceAsyncFn((childSpan) => Promise.all([
  69. // Parse from remote hosts & domain lists
  70. hostsDownloads.map(task => task(childSpan).then(appendArrayToRejectOutput)),
  71. hostsExtraDownloads.map(task => task(childSpan).then(appendArrayToRejectExtraOutput)),
  72. domainListsDownloads.map(task => task(childSpan).then(appendArrayToRejectOutput)),
  73. domainListsExtraDownloads.map(task => task(childSpan).then(appendArrayToRejectExtraOutput)),
  74. adguardFiltersDownloads.map(
  75. task => task(childSpan).then(({ whiteDomains, whiteDomainSuffixes, blackDomains, blackDomainSuffixes }) => {
  76. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomains);
  77. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomainSuffixes, suffix => '.' + suffix);
  78. rejectOutput.bulkAddDomain(blackDomains);
  79. rejectOutput.bulkAddDomainSuffix(blackDomainSuffixes);
  80. })
  81. ),
  82. adguardFiltersExtraDownloads.map(
  83. task => task(childSpan).then(({ whiteDomains, whiteDomainSuffixes, blackDomains, blackDomainSuffixes }) => {
  84. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomains);
  85. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomainSuffixes, suffix => '.' + suffix);
  86. rejectOutput.bulkAddDomain(blackDomains);
  87. rejectOutput.bulkAddDomainSuffix(blackDomainSuffixes);
  88. })
  89. ),
  90. adguardFiltersWhitelistsDownloads.map(
  91. task => task(childSpan).then(({ whiteDomains, whiteDomainSuffixes, blackDomains, blackDomainSuffixes }) => {
  92. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomains);
  93. addArrayElementsToSet(filterRuleWhitelistDomainSets, whiteDomainSuffixes, suffix => '.' + suffix);
  94. addArrayElementsToSet(filterRuleWhitelistDomainSets, blackDomains);
  95. addArrayElementsToSet(filterRuleWhitelistDomainSets, blackDomainSuffixes, suffix => '.' + suffix);
  96. })
  97. ),
  98. getPhishingDomains(childSpan).then(appendArrayToRejectExtraOutput),
  99. readLocalRejectDomainsetPromise.then(appendArrayToRejectOutput),
  100. readLocalRejectDomainsetPromise.then(appendArrayToRejectExtraOutput),
  101. readLocalRejectExtraDomainsetPromise.then(appendArrayToRejectExtraOutput),
  102. // Dedupe domainSets
  103. // span.traceChildAsync('collect black keywords/suffixes', async () =>
  104. /**
  105. * Collect DOMAIN, DOMAIN-SUFFIX, and DOMAIN-KEYWORD from non_ip/reject.conf for deduplication
  106. * DOMAIN-WILDCARD is not really useful for deduplication, it is only included in AdGuardHome output
  107. */
  108. rejectOutput.addFromRuleset(readLocalRejectRulesetPromise),
  109. rejectExtraOutput.addFromRuleset(readLocalRejectRulesetPromise)
  110. ].flat()));
  111. if (foundDebugDomain.value) {
  112. // eslint-disable-next-line sukka/unicorn/no-process-exit -- cli App
  113. process.exit(1);
  114. }
  115. await Promise.all([
  116. rejectOutput.done(),
  117. rejectExtraOutput.done()
  118. ]);
  119. // whitelist
  120. span.traceChildSync('whitelist', () => {
  121. for (const domain of filterRuleWhitelistDomainSets) {
  122. rejectOutput.whitelistDomain(domain);
  123. rejectExtraOutput.whitelistDomain(domain);
  124. }
  125. rejectOutput.domainTrie.dump(rejectExtraOutput.whitelistDomain.bind(rejectExtraOutput));
  126. });
  127. await Promise.all([
  128. rejectOutput.write(),
  129. rejectExtraOutput.write()
  130. ]);
  131. // we are going to re-use rejectOutput's domainTrie and mutate it
  132. // so we must wait until we write rejectOutput to disk after we can mutate its trie
  133. const rejectOutputAdGuardHome = new FileOutput(span, 'reject-adguardhome')
  134. .withTitle('Sukka\'s Ruleset - Blocklist for AdGuardHome')
  135. .withDescription([
  136. 'The domainset supports AD blocking, tracking protection, privacy protection, anti-phishing, anti-mining'
  137. ])
  138. .replaceStrategies([new AdGuardHome(OUTPUT_INTERNAL_DIR)]);
  139. rejectOutputAdGuardHome.domainTrie = rejectOutput.domainTrie;
  140. await rejectOutputAdGuardHome
  141. .addFromRuleset(readLocalMyRejectRulesetPromise)
  142. .addFromRuleset(readLocalRejectRulesetPromise)
  143. .addFromRuleset(readLocalRejectDropRulesetPromise)
  144. .addFromRuleset(readLocalRejectNoDropRulesetPromise)
  145. .write();
  146. });