reject-data-source.ts 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229
  1. import { TTL } from './cache-filesystem';
  2. export const HOSTS = [
  3. ['https://pgl.yoyo.org/adservers/serverlist.php?hostformat=hosts&showintro=0&mimetype=plaintext', true],
  4. ['https://someonewhocares.org/hosts/hosts', true],
  5. // no coin list is not actively maintained, but it updates daily when being maintained, so we set a 3 days cache ttl
  6. ['https://raw.githubusercontent.com/hoshsadiq/adblock-nocoin-list/master/hosts.txt', false, false, TTL.THREE_DAYS()],
  7. // have not been updated for more than a year, so we set a 14 days cache ttl
  8. ['https://raw.githubusercontent.com/crazy-max/WindowsSpyBlocker/master/data/hosts/spy.txt', true, false, TTL.TWO_WEEKS()],
  9. ['https://raw.githubusercontent.com/jerryn70/GoodbyeAds/master/Extension/GoodbyeAds-Xiaomi-Extension.txt', false],
  10. ['https://raw.githubusercontent.com/jerryn70/GoodbyeAds/master/Extension/GoodbyeAds-Huawei-Extension.txt', false],
  11. // ad-wars is not actively maintained, so we set a 7 days cache ttl
  12. ['https://raw.githubusercontent.com/jdlingyu/ad-wars/master/hosts', false, false, TTL.ONE_WEEK()],
  13. ['https://raw.githubusercontent.com/durablenapkin/block/master/luminati.txt', true],
  14. // CoinBlockerList
  15. // Although the hosts file is still actively maintained, the hosts_browser file is not updated since 2021-07, so we set a 14 days cache ttl
  16. ['https://zerodot1.gitlab.io/CoinBlockerLists/hosts_browser', true, true, TTL.TWO_WEEKS()],
  17. // Curben's UrlHaus Malicious URL Blocklist
  18. // 'https://curbengh.github.io/urlhaus-filter/urlhaus-filter-agh-online.txt',
  19. // 'https://urlhaus-filter.pages.dev/urlhaus-filter-agh-online.txt',
  20. ['https://curbengh.github.io/urlhaus-filter/urlhaus-filter-hosts.txt', true, true],
  21. // Curben's Phishing URL Blocklist
  22. // Covered by lib/get-phishing-domains.ts
  23. // 'https://curbengh.github.io/phishing-filter/phishing-filter-agh.txt'
  24. // 'https://phishing-filter.pages.dev/phishing-filter-agh.txt'
  25. // ['https://curbengh.github.io/phishing-filter/phishing-filter-hosts.txt', true, true],
  26. // Curben's PUP Domains Blocklist
  27. // 'https://curbengh.github.io/pup-filter/pup-filter-agh.txt'
  28. // 'https://pup-filter.pages.dev/pup-filter-agh.txt'
  29. // The PUP filter has paused the update since 2023-05, so we set a 14 days cache ttl
  30. ['https://curbengh.github.io/pup-filter/pup-filter-hosts.txt', true, true, TTL.TWO_WEEKS()]
  31. ] as const;
  32. export const DOMAIN_LISTS = [
  33. // BarbBlock
  34. // The barbblock list has never been updated since 2019-05, so we set a 14 days cache ttl
  35. ['https://paulgb.github.io/BarbBlock/blacklists/domain-list.txt', true, TTL.TWO_WEEKS()],
  36. // DigitalSide Threat-Intel - OSINT Hub
  37. // Update once per day
  38. ['https://osint.digitalside.it/Threat-Intel/lists/latestdomains.txt', true, 24 * 60 * 60 * 1000],
  39. // AdGuard CNAME Filter Combined
  40. // Update on a 7 days basis, so we add a 3 hours cache ttl
  41. ['https://raw.githubusercontent.com/AdguardTeam/cname-trackers/master/data/combined_disguised_ads_justdomains.txt', true, TTL.THREE_DAYS()],
  42. ['https://raw.githubusercontent.com/AdguardTeam/cname-trackers/master/data/combined_disguised_trackers_justdomains.txt', true, TTL.THREE_DAYS()],
  43. ['https://raw.githubusercontent.com/AdguardTeam/cname-trackers/master/data/combined_disguised_clickthroughs_justdomains.txt', true, TTL.THREE_DAYS()],
  44. ['https://raw.githubusercontent.com/AdguardTeam/cname-trackers/master/data/combined_disguised_microsites_justdomains.txt', true, TTL.THREE_DAYS()],
  45. ['https://raw.githubusercontent.com/AdguardTeam/cname-trackers/master/data/combined_disguised_mail_trackers_justdomains.txt', true, TTL.THREE_DAYS()]
  46. ] as const;
  47. export const ADGUARD_FILTERS = [
  48. // EasyList
  49. [
  50. 'https://easylist.to/easylist/easylist.txt',
  51. [
  52. 'https://easylist-downloads.adblockplus.org/easylist.txt',
  53. 'https://raw.githubusercontent.com/easylist/easylist/gh-pages/easylist.txt',
  54. 'https://secure.fanboy.co.nz/easylist.txt',
  55. 'https://ublockorigin.github.io/uAssetsCDN/thirdparties/easylist.txt',
  56. 'https://ublockorigin.pages.dev/thirdparties/easylist.txt'
  57. ],
  58. 12 * 60 * 60 * 1000
  59. ],
  60. // EasyPrivacy
  61. [
  62. 'https://easylist.to/easylist/easyprivacy.txt',
  63. [
  64. 'https://secure.fanboy.co.nz/easyprivacy.txt',
  65. 'https://raw.githubusercontent.com/easylist/easylist/gh-pages/easyprivacy.txt',
  66. 'https://easylist-downloads.adblockplus.org/easyprivacy.txt',
  67. 'https://ublockorigin.github.io/uAssetsCDN/thirdparties/easyprivacy.txt',
  68. 'https://ublockorigin.pages.dev/thirdparties/easyprivacy.txt'
  69. ],
  70. 12 * 60 * 60 * 1000
  71. ],
  72. // AdGuard DNS Filter
  73. [
  74. 'https://adguardteam.github.io/AdGuardSDNSFilter/Filters/filter.txt',
  75. [
  76. 'https://filters.adtidy.org/extension/ublock/filters/15_optimized.txt',
  77. 'https://adguardteam.github.io/HostlistsRegistry/assets/filter_1.txt'
  78. ]
  79. ],
  80. // uBlock Origin Filter List
  81. [
  82. 'https://ublockorigin.github.io/uAssetsCDN/filters/filters.min.txt',
  83. [
  84. 'https://ublockorigin.pages.dev/filters/filters.min.txt'
  85. ]
  86. ],
  87. // uBlock Origin Badware Risk List
  88. [
  89. 'https://ublockorigin.github.io/uAssetsCDN/filters/badware.min.txt',
  90. [
  91. 'https://ublockorigin.pages.dev/filters/badware.min.txt'
  92. ]
  93. ],
  94. // uBlock Origin Privacy List
  95. [
  96. 'https://ublockorigin.github.io/uAssetsCDN/filters/privacy.min.txt',
  97. [
  98. 'https://ublockorigin.pages.dev/filters/privacy.min.txt'
  99. ]
  100. ],
  101. // uBlock Origin Resource Abuse: merged in uBlock Origin Privacy List
  102. // [
  103. // 'https://ublockorigin.github.io/uAssetsCDN/filters/resource-abuse.txt',
  104. // [
  105. // 'https://ublockorigin.pages.dev/filters/resource-abuse.txt'
  106. // ]
  107. // ],
  108. // uBlock Origin Unbreak
  109. [
  110. 'https://ublockorigin.github.io/uAssetsCDN/filters/unbreak.min.txt',
  111. [
  112. 'https://ublockorigin.pages.dev/filters/unbreak.min.txt'
  113. ]
  114. ],
  115. // AdGuard Base Filter
  116. 'https://filters.adtidy.org/extension/ublock/filters/2_without_easylist.txt',
  117. // AdGuard Mobile AD
  118. 'https://filters.adtidy.org/extension/ublock/filters/11_optimized.txt',
  119. // AdGuard Tracking Protection
  120. 'https://filters.adtidy.org/extension/ublock/filters/3_optimized.txt',
  121. // AdGuard Japanese filter
  122. 'https://filters.adtidy.org/extension/ublock/filters/7_optimized.txt',
  123. // AdGuard Chinese filter (EasyList China + AdGuard Chinese filter)
  124. 'https://filters.adtidy.org/extension/ublock/filters/224_optimized.txt',
  125. // AdGuard Annoyances filter
  126. 'https://filters.adtidy.org/android/filters/14_optimized.txt',
  127. // EasyList Germany filter
  128. [
  129. 'https://easylist.to/easylistgermany/easylistgermany.txt',
  130. [
  131. 'https://easylist-downloads.adblockplus.org/easylistgermany.txt'
  132. ]
  133. ],
  134. // GameConsoleAdblockList
  135. 'https://raw.githubusercontent.com/DandelionSprout/adfilt/master/GameConsoleAdblockList.txt',
  136. // PiHoleBlocklist
  137. // Update almost once per 3 months, let's set a 10 days cache ttl
  138. [
  139. 'https://perflyst.github.io/PiHoleBlocklist/SmartTV-AGH.txt',
  140. [
  141. 'https://raw.githubusercontent.com/Perflyst/PiHoleBlocklist/master/SmartTV-AGH.txt'
  142. ],
  143. TTL.TEN_DAYS()
  144. ],
  145. // Spam404
  146. // Not actively maintained, let's use a 10 days cache ttl
  147. ['https://raw.githubusercontent.com/Spam404/lists/master/adblock-list.txt', null, TTL.TEN_DAYS()],
  148. // Brave First Party & First Party CNAME
  149. 'https://raw.githubusercontent.com/brave/adblock-lists/master/brave-lists/brave-firstparty.txt'
  150. ] as const;
  151. export const PREDEFINED_WHITELIST = [
  152. 'localhost',
  153. 'local',
  154. 'localhost.localdomain',
  155. 'broadcasthost',
  156. 'ip6-loopback',
  157. 'ip6-localnet',
  158. 'ip6-mcastprefix',
  159. 'ip6-allnodes',
  160. 'ip6-allrouters',
  161. 'ip6-allhosts',
  162. 'mcastprefix',
  163. 'skk.moe',
  164. 'analytics.google.com',
  165. 'cloud.answerhub.com',
  166. 'ae01.alicdn.com',
  167. 'whoami.akamai.net',
  168. 'whoami.ds.akahelp.net',
  169. 'pxlk9.net.', // This one is malformed from EasyList, which I will manually add instead
  170. 'instant.page', // No, it doesn't violate anyone's privacy. I will whitelist it
  171. 'piwik.pro',
  172. 'mixpanel.com',
  173. 'cdn.mxpnl.com',
  174. 'heapanalytics.com',
  175. 'segment.com',
  176. 'segmentify.com',
  177. 't.co', // pgl yoyo add t.co to the blacklist
  178. 'survicate.com', // AdGuardDNSFilter
  179. 'perfops.io', // AdGuardDNSFilter
  180. 'd2axgrpnciinw7.cloudfront.net', // ADGuardDNSFilter
  181. 'tb-lb.sb-cd.com', // AdGuard
  182. 'storage.yandexcloud.net', // phishing list
  183. 'login.microsoftonline.com', // phishing list
  184. 'api.xiaomi.com', // https://github.com/jerryn70/GoodbyeAds/issues/281
  185. 'api.io.mi.com', // https://github.com/jerryn70/GoodbyeAds/issues/281
  186. 'cdn.userreport.com', // https://github.com/AdguardTeam/AdGuardSDNSFilter/issues/1158
  187. 'ip-api.com',
  188. 'fastly-analytics.com',
  189. 'syd1.digitaloceanspaces.com',
  190. 's3.nl-ams.scw.cloud',
  191. 'geolocation-db.com',
  192. 'uploads.codesandbox.io',
  193. // Affect Windows ISO download
  194. // https://raw.githubusercontent.com/AdguardTeam/cname-trackers/master/data/combined_disguised_trackers.txt
  195. 'vlscppe.microsoft.com',
  196. // OpenAI use this for A/B testing
  197. // Fuck Peter Lowe Hosts
  198. 'pstmrk.it'
  199. ];
  200. export const PREDEFINED_ENFORCED_BACKLIST = [
  201. 'telemetry.mozilla.org'
  202. ];
  203. export const PREDEFINED_ENFORCED_WHITELIST = [
  204. 'godaddysites.com',
  205. 'web.app',
  206. 'firebaseapp.com',
  207. 'ipfs.nftstorage.link',
  208. 'ipfs.4everland.io',
  209. 'ipfs.cf-ipfs.com',
  210. 'ipfs.dweb.link',
  211. 'ipfs.infura-ipfs.io',
  212. 'ipfs.fleek.cool',
  213. 'repl.co',
  214. 'w3s.link',
  215. 'translate.goog',
  216. 'backblazeb2.com',
  217. 'workers.dev',
  218. 'r2.dev',
  219. 'glitch.me',
  220. 'netlify.app',
  221. 'blogspot.com',
  222. 'appspot.com'
  223. ];